nfsdctl: add support for new lockd configuration interface
Subsystems (2)
Jeff Layton
maintainerv2 cover nfsdctl: add support for new lockd configuration interface
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698
There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
---
Changes in v2:
- properly regenerate manpages
- fix up bogus merge conflict
- add D_GENERAL xlog messages when nfsdctl starts and exits
- Link to v1: https://lore.kernel.org/r/[email protected]
---
Jeff Layton (3):
nfsdctl: convert to xlog()
nfsdctl: fix the --version option
nfsdctl: add necessary bits to configure lockd
configure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++ utils/nfsdctl/nfsdctl.8 | 15 +- utils/nfsdctl/nfsdctl.adoc | 8 + utils/nfsdctl/nfsdctl.c | 331 ++++++++++++++++++++++++++++++++++-------- 5 files changed, 321 insertions(+), 66 deletions(-)
---
base-commit: 65f4cc3a6ce1472ee4092c4bbf4b19beb0a8217b
change-id: 20250109-lockd-nl-6272fa9e8a5d
Best regards,
Signature · 1 line
Jeff Layton <[email protected]>
Jeff Layton
maintainerPATCH v2 1/3 nfsdctl: convert to xlog()
Convert all of the fprintf(stderr, ...) calls to xlog(L_ERROR, ...) calls. Change the -d option to not take an argument, and add a -s option that will make nfsdctl log to syslog instead of stderr.
Also remove the extraneous error message in run_commandline, and add a couple of trivial debug messages when the program starts and ends
--- utils/nfsdctl/nfsdctl.8 | 9 +++- utils/nfsdctl/nfsdctl.adoc | 3 ++ utils/nfsdctl/nfsdctl.c | 111 +++++++++++++++++++++++++-------------------- 3 files changed, 71 insertions(+), 52 deletions(-)
utils/nfsdctl/nfsdctl.8 +7 -2
@@ -2,12 +2,12 @@ .\" Title: nfsdctl .\" Author: Jeff Layton .\" Generator: Asciidoctor 2.0.20 -.\" Date: 2024-12-30 +.\" Date: 2025-01-09 .\" Manual: \ \& .\" Source: \ \& .\" Language: English .\" -.TH "NFSDCTL" "8" "2024-12-30" "\ \&" "\ \&" +.TH "NFSDCTL" "8" "2025-01-09" "\ \&" "\ \&" .ie \n(.g .ds Aq \(aq .el .ds Aq ' .ss \n[.ss] 0
@@ -60,6 +60,11 @@ Enable debug logging Print program help text .RE .sp +\fB\-s, \-\-syslog\fP +.RS 4 +Log to syslog instead of to stderr (the default) +.RE +.sp \fB\-V, \-\-version\fP .RS 4 Print program version
utils/nfsdctl/nfsdctl.adoc +3 -0
@@ -30,6 +30,9 @@ To get information about different subcommand usage, pass the subcommand the *-h, --help*:: Print program help text +*-s, --syslog*:: + Log to syslog instead of to stderr (the default) + *-V, --version*:: Print program version
utils/nfsdctl/nfsdctl.c +61 -50
369 lines not on this page · raw message
Signature · 1 line
2.47.1
Jeff Layton
maintainerPATCH v2 2/3 nfsdctl: fix the --version option
In reply to Jeff Layton--- utils/nfsdctl/nfsdctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
utils/nfsdctl/nfsdctl.c +1 -1
@@ -1550,7 +1550,7 @@ int main(int argc, char **argv)xlog_stderr (0 );break ;case 'V' : -// FIXME: print_version(); +fprintf (stdout,"nfsdctl: " VERSION "\n" );return 0 ; } }
Signature · 1 line
2.47.1
Jeff Layton
maintainerPATCH v2 3/3 nfsdctl: add necessary bits to configure lockd
In reply to Jeff LaytonThe legacy rpc.nfsd program would configure the nlm_grace_period to match the NFSv4 grace period when starting the server. This adds similar functionality to the autostart subcommand using the new lockd netlink configuration interfaces. In addition, if lockd's udp or tcp listener ports are specified, also configure them before starting the server.
A "nlm" subcommand is also added that will display the current lockd config settings in the current net ns. In the future, we could add the ability to set these values using the "nlm" subcommand, but for now I didn't see a real use-case for that, so I left it out.
--- configure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++++ utils/nfsdctl/nfsdctl.8 | 6 ++ utils/nfsdctl/nfsdctl.adoc | 5 + utils/nfsdctl/nfsdctl.c | 218 +++++++++++++++++++++++++++++++++++++++--- 5 files changed, 249 insertions(+), 13 deletions(-)
configure.ac +4 -0
@@ -268,6 +268,10 @@ AC_ARG_ENABLE(nfsdctl, [[int foo = NFSD_CMD_POOL_MODE_GET;]])], [AC_DEFINE([USE_SYSTEM_NFSD_NETLINK_H], 1, ["Use system's linux/nfsd_netlink.h"])]) + AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[#include <linux/lockd_netlink.h>]], + [[int foo = LOCKD_CMD_SERVER_GET;]])], + [AC_DEFINE([USE_SYSTEM_LOCKD_NETLINK_H], 1, + ["Use system's linux/lockd_netlink.h"])]) fi AC_ARG_ENABLE(nfsv4server,
utils/nfsdctl/lockd_netlink.h new file +29 -0
@@ -0,0 +1,29 @@ +/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */ +/* Do not edit directly, auto-generated from: */ +/* Documentation/netlink/specs/lockd.yaml */ +/* YNL-GEN uapi header */ + +#ifndef _UAPI_LINUX_LOCKD_NETLINK_H +#define _UAPI_LINUX_LOCKD_NETLINK_H + +#define LOCKD_FAMILY_NAME "lockd" +#define LOCKD_FAMILY_VERSION 1 + +enum { +LOCKD_A_SERVER_GRACETIME =1 , +LOCKD_A_SERVER_TCP_PORT , +LOCKD_A_SERVER_UDP_PORT , + + __LOCKD_A_SERVER_MAX, +LOCKD_A_SERVER_MAX = (__LOCKD_A_SERVER_MAX -1 ) +}; + +enum { +LOCKD_CMD_SERVER_SET =1 , +LOCKD_CMD_SERVER_GET , + + __LOCKD_CMD_MAX, +LOCKD_CMD_MAX = (__LOCKD_CMD_MAX -1 ) +}; + +#endif /* _UAPI_LINUX_LOCKD_NETLINK_H */
utils/nfsdctl/nfsdctl.8 +6 -0
@@ -127,6 +127,12 @@ colon separated form, and must be enclosed in square brackets. .if n .RE .RE .sp +\fBnlm\fP +.RS 4 +Get information about NLM (lockd) settings in the current net namespace. This +subcommand takes no arguments. +.RE +.sp \fBstatus\fP .RS 4 Get information about RPCs currently executing in the server. This subcommand
utils/nfsdctl/nfsdctl.adoc +5 -0
@@ -67,6 +67,11 @@ Each subcommand can also accept its own set of options and arguments. The addresses must be in dotted-quad form. IPv6 addresses should be in standard colon separated form, and must be enclosed in square brackets. +*nlm*:: + + Get information about NLM (lockd) settings in the current net namespace. This + subcommand takes no arguments. + *status*:: Get information about RPCs currently executing in the server. This subcommand
utils/nfsdctl/nfsdctl.c +205 -13
347 lines not on this page · raw message
Signature · 1 line
2.47.1
Tom Talpey
reviewerRe: [PATCH v2 3/3] nfsdctl: add necessary bits to configure lockd
Quoted 10 lines · Jeff Layton
The legacy rpc.nfsd program would configure the
nlm_grace_periodto match the NFSv4 grace period when starting the server. This adds similar functionality to the autostart subcommand using the new lockd netlink configuration interfaces. In addition, if lockd's udp or tcp listener ports are specified, also configure them before starting the server.A "nlm" subcommand is also added that will display the current lockd config settings in the current net ns. In the future, we could add the ability to set these values using the "nlm" subcommand, but for now I didn't see a real use-case for that, so I left it out.
It seems unnatural that the nlm_grace_period is tied to the netns.
It seems to me it's more dependent on the network and its likely failure modes, the backend storage/filesystem, and perhaps the scale of clients performing possibly-conflicting locks. Oh, and also perhaps the minor version, since 4.1+ have the RECLAIM_COMPLETE termination event.
Food for thought, perhaps.
Tom.
Quoted 466 lines
Signed-off-by Jeff Layton <[email protected]>--- configure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++++ utils/nfsdctl/nfsdctl.8 | 6 ++ utils/nfsdctl/nfsdctl.adoc | 5 + utils/nfsdctl/nfsdctl.c | 218 +++++++++++++++++++++++++++++++++++++++--- 5 files changed, 249 insertions(+), 13 deletions(-)configure.ac +4
-0@@ -268,6 +268,10 @@ AC_ARG_ENABLE(nfsdctl, [[int foo = NFSD_CMD_POOL_MODE_GET;]])], [AC_DEFINE([USE_SYSTEM_NFSD_NETLINK_H], 1, ["Use system's linux/nfsd_netlink.h"])]) + AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[#include <linux/lockd_netlink.h>]], + [[int foo = LOCKD_CMD_SERVER_GET;]])], + [AC_DEFINE([USE_SYSTEM_LOCKD_NETLINK_H], 1, + ["Use system's linux/lockd_netlink.h"])]) fi AC_ARG_ENABLE(nfsv4server,utils/nfsdctl/lockd_netlink.h new file +29
-0@@ -0,0 +1,29 @@ +/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */ +/* Do not edit directly, auto-generated from: */ +/* Documentation/netlink/specs/lockd.yaml */ +/* YNL-GEN uapi header */ + +#ifndef _UAPI_LINUX_LOCKD_NETLINK_H +#define _UAPI_LINUX_LOCKD_NETLINK_H + +#define LOCKD_FAMILY_NAME "lockd" +#define LOCKD_FAMILY_VERSION 1 + +enum { +LOCKD_A_SERVER_GRACETIME =1 , +LOCKD_A_SERVER_TCP_PORT , +LOCKD_A_SERVER_UDP_PORT , + + __LOCKD_A_SERVER_MAX, +LOCKD_A_SERVER_MAX = (__LOCKD_A_SERVER_MAX -1 ) +}; + +enum { +LOCKD_CMD_SERVER_SET =1 , +LOCKD_CMD_SERVER_GET , + + __LOCKD_CMD_MAX, +LOCKD_CMD_MAX = (__LOCKD_CMD_MAX -1 ) +}; + +#endif /* _UAPI_LINUX_LOCKD_NETLINK_H */ utils/nfsdctl/nfsdctl.8 +6
-0@@ -127,6 +127,12 @@ colon separated form, and must be enclosed in square brackets. .if n .RE .RE .sp +\fBnlm\fP +.RS 4 +Get information about NLM (lockd) settings in the current net namespace. This +subcommand takes no arguments. +.RE +.sp \fBstatus\fP .RS 4 Get information about RPCs currently executing in the server. This subcommandutils/nfsdctl/nfsdctl.adoc +5
-0@@ -67,6 +67,11 @@ Each subcommand can also accept its own set of options and arguments. The addresses must be in dotted-quad form. IPv6 addresses should be in standard colon separated form, and must be enclosed in square brackets. +*nlm*:: + + Get information about NLM (lockd) settings in the current net namespace. This + subcommand takes no arguments. + *status*:: Get information about RPCs currently executing in the server. This subcommandutils/nfsdctl/nfsdctl.c +205
-13@@ -35,6 +35,12 @@#include "nfsd_netlink.h" #endif +#ifdef USE_SYSTEM_LOCKD_NETLINK_H +#include <linux/lockd_netlink.h> +#else +#include "lockd_netlink.h" +#endif +#include "nfsdctl.h" #include "conffile.h" #include "xlog.h" @@ -348,6 +354,28 @@ static void parse_pool_mode_get(struct genlmsghdr *gnlh) } } +static void parse_lockd_get (struct genlmsghdr *gnlh) +{ +struct nlattr *attr; +int rem; + +nla_for_each_attr (attr,genlmsg_attrdata (gnlh,0 ), +genlmsg_attrlen (gnlh,0 ), rem) { +switch (nla_type (attr)) { +case LOCKD_A_SERVER_GRACETIME : +printf ("gracetime: %u\n" ,nla_get_u32 (attr)); +break ; +case LOCKD_A_SERVER_TCP_PORT : +printf ("tcp_port: %hu\n" ,nla_get_u16 (attr)); +break ; +case LOCKD_A_SERVER_UDP_PORT : +printf ("udp_port: %hu\n" ,nla_get_u16 (attr)); +break ; +default : +break ; + } + } +}static int recv_handler (struct nl_msg *msg,void *arg) {struct genlmsghdr *gnlh =nlmsg_data (nlmsg_hdr (msg));@@ -368,6 +396,9 @@ static int recv_handler(struct nl_msg *msg, void *arg)case NFSD_CMD_POOL_MODE_GET :parse_pool_mode_get (gnlh);break ; +case LOCKD_CMD_SERVER_GET : +parse_lockd_get (gnlh); +break ;default :break ; }@@ -398,12 +429,12 @@ static struct nl_sock *netlink_sock_alloc(void)return sock; } -static struct nl_msg *netlink_msg_alloc (struct nl_sock *sock) +static struct nl_msg *netlink_msg_alloc (struct nl_sock *sock,const char *family) {struct nl_msg *msg;int id; - id =genl_ctrl_resolve (sock,NFSD_FAMILY_NAME ); + id =genl_ctrl_resolve (sock, family);if (id <0 ) {xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME );return NULL ;@@ -447,7 +478,7 @@ static int status_func(struct nl_sock *sock, int argc, char ** argv) } } - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -495,7 +526,7 @@ static int threads_doit(struct nl_sock *sock, int cmd, int grace, int lease,struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -607,7 +638,7 @@ static int fetch_nfsd_versions(struct nl_sock *sock)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -672,7 +703,7 @@ static int set_nfsd_versions(struct nl_sock *sock)struct nl_cb *cb;int i, ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -825,7 +856,7 @@ static int fetch_current_listeners(struct nl_sock *sock)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1054,7 +1085,7 @@ static int set_listeners(struct nl_sock *sock)struct nl_cb *cb;int i, ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1170,7 +1201,7 @@ static int pool_mode_doit(struct nl_sock *sock, int cmd, const char *pool_mode)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1249,6 +1280,131 @@ static int pool_mode_func(struct nl_sock *sock, int argc, char **argv)return pool_mode_doit (sock, cmd, pool_mode); } +static int lockd_config_doit (struct nl_sock *sock,int cmd,int grace,int tcpport,int udpport) +{ +struct genlmsghdr *ghdr; +struct nlmsghdr *nlh; +struct nl_msg *msg; +struct nl_cb *cb; +int ret; + +if (cmd ==LOCKD_CMD_SERVER_SET ) { +/* Do nothing if there is nothing to set */ +if (!grace && !tcpport && !udpport) +return 0 ; + } + + msg =netlink_msg_alloc (sock,LOCKD_FAMILY_NAME ); +if (!msg) +return 1 ; + + nlh =nlmsg_hdr (msg); +if (cmd ==LOCKD_CMD_SERVER_SET ) { +if (grace) +nla_put_u32 (msg,LOCKD_A_SERVER_GRACETIME , grace); +if (tcpport) +nla_put_u16 (msg,LOCKD_A_SERVER_TCP_PORT , tcpport); +if (udpport) +nla_put_u16 (msg,LOCKD_A_SERVER_UDP_PORT , udpport); + } + + ghdr =nlmsg_data (nlh); + ghdr->cmd = cmd; + + cb =nl_cb_alloc (NL_CB_CUSTOM ); +if (!cb) { +xlog (L_ERROR ,"failed to allocate netlink callbacks\n" ); + ret =1 ; + goto out; + } + + ret =nl_send_auto (sock, msg); +if (ret <0 ) { +xlog (L_ERROR ,"send failed (%d)!\n" , ret); + goto out_cb; + } + + ret =1 ; +nl_cb_err (cb,NL_CB_CUSTOM , error_handler, &ret); +nl_cb_set (cb,NL_CB_FINISH ,NL_CB_CUSTOM , finish_handler, &ret); +nl_cb_set (cb,NL_CB_ACK ,NL_CB_CUSTOM , ack_handler, &ret); +nl_cb_set (cb,NL_CB_VALID ,NL_CB_CUSTOM , recv_handler,NULL ); + +while (ret >0 ) +nl_recvmsgs (sock, cb); +if (ret <0 ) { +xlog (L_ERROR ,"Error: %s\n" ,strerror (-ret)); + ret =1 ; + } +out_cb: +nl_cb_put (cb); +out: +nlmsg_free (msg); +return ret; +} + +static int get_service (const char *svc) +{ +struct addrinfo *res, hints = { .ai_flags =AI_PASSIVE }; +int ret, port; + +if (!svc) +return 0 ; + + ret =getaddrinfo (NULL , svc, &hints, &res); +if (ret) { +xlog (L_ERROR ,"getaddrinfo of \"%s\" failed: %s\n" , + svc,gai_strerror (ret)); +return -1 ; + } + +switch (res->ai_family) { +case AF_INET : + { +struct sockaddr_in *sin = (struct sockaddr_in *)res->ai_addr; + + port =ntohs (sin->sin_port); + } +break ; +case AF_INET6 : + { +struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)res->ai_addr; + + port =ntohs (sin6->sin6_port); + } +break ; +default : +xlog (L_ERROR ,"Bad address family: %d\n" , res->ai_family); + port =-1 ; + } +freeaddrinfo (res); +return port; +} + +static int lockd_configure (struct nl_sock *sock,int grace) +{ +char *tcp_svc, *udp_svc; +int tcpport =0 , udpport =0 ; +int ret; + + tcp_svc =conf_get_str ("lockd" ,"port" ); +if (tcp_svc) { + tcpport =get_service (tcp_svc); +if (tcpport <0 ) +return 1 ; + } + + udp_svc =conf_get_str ("lockd" ,"udp-port" ); +if (udp_svc) { + udpport =get_service (udp_svc); +if (udpport <0 ) +return 1 ; + } + +return lockd_config_doit (sock,LOCKD_CMD_SERVER_SET , grace, tcpport, udpport); +} + +#define MAX_LISTENER_LEN (64 * 2 + 16) static void@@ -1355,6 +1511,13 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv)read_nfsd_conf (); + grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); + ret =lockd_configure (sock, grace); +if (ret) { +xlog (L_ERROR ,"lockd configuration failure" ); +return ret; + } + pool_mode =conf_get_str ("nfsd" ,"pool-mode" );if (pool_mode) { ret =pool_mode_doit (sock,NFSD_CMD_POOL_MODE_SET , pool_mode);@@ -1370,15 +1533,12 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv)if (ret)return ret; +xlog (D_GENERAL ,"configuring listeners" );configure_listeners (); ret =set_listeners (sock);if (ret)return ret; - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); - lease =conf_get_num ("nfsd" ,"lease-time" ,0 ); - scope =conf_get_str ("nfsd" ,"scope" ); - thread_str =conf_get_list ("nfsd" ,"threads" ); pools = thread_str ? thread_str->cnt :1 ;@@ -1402,6 +1562,9 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) threads[0 ] =DEFAULT_AUTOSTART_THREADS ; } + lease =conf_get_num ("nfsd" ,"lease-time" ,0 ); + scope =conf_get_str ("nfsd" ,"scope" ); + ret =threads_doit (sock,NFSD_CMD_THREADS_SET , grace, lease, pools, threads, scope); out:@@ -1409,6 +1572,30 @@ out:return ret; } +static void nlm_usage (void ) +{ +printf ("Usage: %s nlm\n" , taskname); +printf (" Get the current settings for the NLM (lockd) server.\n" ); +} + +static int nlm_func (struct nl_sock *sock,int argc,char ** argv) +{ +int *threads, grace, lease, idx, ret, opt, pools; +struct conf_list *thread_str; +struct conf_list_node *n; +char *scope, *pool_mode; + + optind =1 ; +while ((opt =getopt_long (argc, argv,"h" , help_only_options,NULL )) !=-1 ) { +switch (opt) { +case 'h' : +nlm_usage (); +return 0 ; + } + } +return lockd_config_doit (sock,LOCKD_CMD_SERVER_GET ,0 ,0 ,0 ); +} +enum nfsdctl_commands {NFSDCTL_STATUS ,NFSDCTL_THREADS ,@@ -1416,6 +1603,7 @@ enum nfsdctl_commands {NFSDCTL_LISTENER ,NFSDCTL_AUTOSTART ,NFSDCTL_POOL_MODE , + NFSDCTL_NLM, };static int parse_command (char *str)@@ -1432,6 +1620,8 @@ static int parse_command(char *str)return NFSDCTL_AUTOSTART ;if (!strcmp (str,"pool-mode" ))return NFSDCTL_POOL_MODE ; +if (!strcmp (str,"nlm" )) +return NFSDCTL_NLM ;return -1 ; }@@ -1444,6 +1634,7 @@ static nfsdctl_func func[] = { [NFSDCTL_LISTENER] = listener_func, [NFSDCTL_AUTOSTART] = autostart_func, [NFSDCTL_POOL_MODE] = pool_mode_func, + [NFSDCTL_NLM] = nlm_func, };static void usage (void )@@ -1460,6 +1651,7 @@ static void usage(void)printf (" listener get/set listener info\n" );printf (" version get/set supported NFS versions\n" );printf (" threads get/set nfsd thread settings\n" ); +printf (" nlm get current nlm settings\n" );printf (" status get current RPC processing info\n" );printf (" autostart start server with settings from /etc/nfs.conf\n" ); }
Jeff Layton
maintainerRe: [PATCH v2 3/3] nfsdctl: add necessary bits to configure lockd
Quoted 20 lines · Tom Talpey
Quoted 10 lines · Jeff Layton
The legacy rpc.nfsd program would configure the
nlm_grace_periodto match the NFSv4 grace period when starting the server. This adds similar functionality to the autostart subcommand using the new lockd netlink configuration interfaces. In addition, if lockd's udp or tcp listener ports are specified, also configure them before starting the server.A "nlm" subcommand is also added that will display the current lockd config settings in the current net ns. In the future, we could add the ability to set these values using the "nlm" subcommand, but for now I didn't see a real use-case for that, so I left it out.
It seems unnatural that the
nlm_grace_periodis tied to the netns.It seems to me it's more dependent on the network and its likely failure modes, the backend storage/filesystem, and perhaps the scale of clients performing possibly-conflicting locks. Oh, and also perhaps the minor version, since 4.1+ have the
RECLAIM_COMPLETEtermination event.
Quoted 4 lines
Food for thought, perhaps.
Tom.
Fair point. More food:
My guess is that nlm_grace_period handling is likely horribly broken in multi-container scenarios anyway. If you have multiple nfs server containers with different grace period settings, then they will all be competing to set the global nlm_grace_period.
Most clients end up reclaiming quickly enough that we don't hit major problems here, but it would be good to make this more "airtight".
Quoted 467 lines
Quoted 466 lines
Signed-off-by Jeff Layton <[email protected]>--- configure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++++ utils/nfsdctl/nfsdctl.8 | 6 ++ utils/nfsdctl/nfsdctl.adoc | 5 + utils/nfsdctl/nfsdctl.c | 218 +++++++++++++++++++++++++++++++++++++++--- 5 files changed, 249 insertions(+), 13 deletions(-)configure.ac +4
-0@@ -268,6 +268,10 @@ AC_ARG_ENABLE(nfsdctl, [[int foo = NFSD_CMD_POOL_MODE_GET;]])], [AC_DEFINE([USE_SYSTEM_NFSD_NETLINK_H], 1, ["Use system's linux/nfsd_netlink.h"])]) + AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[#include <linux/lockd_netlink.h>]], + [[int foo = LOCKD_CMD_SERVER_GET;]])], + [AC_DEFINE([USE_SYSTEM_LOCKD_NETLINK_H], 1, + ["Use system's linux/lockd_netlink.h"])]) fi AC_ARG_ENABLE(nfsv4server,utils/nfsdctl/lockd_netlink.h new file +29
-0@@ -0,0 +1,29 @@ +/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */ +/* Do not edit directly, auto-generated from: */ +/* Documentation/netlink/specs/lockd.yaml */ +/* YNL-GEN uapi header */ + +#ifndef _UAPI_LINUX_LOCKD_NETLINK_H +#define _UAPI_LINUX_LOCKD_NETLINK_H + +#define LOCKD_FAMILY_NAME "lockd" +#define LOCKD_FAMILY_VERSION 1 + +enum { +LOCKD_A_SERVER_GRACETIME =1 , +LOCKD_A_SERVER_TCP_PORT , +LOCKD_A_SERVER_UDP_PORT , + + __LOCKD_A_SERVER_MAX, +LOCKD_A_SERVER_MAX = (__LOCKD_A_SERVER_MAX -1 ) +}; + +enum { +LOCKD_CMD_SERVER_SET =1 , +LOCKD_CMD_SERVER_GET , + + __LOCKD_CMD_MAX, +LOCKD_CMD_MAX = (__LOCKD_CMD_MAX -1 ) +}; + +#endif /* _UAPI_LINUX_LOCKD_NETLINK_H */ utils/nfsdctl/nfsdctl.8 +6
-0@@ -127,6 +127,12 @@ colon separated form, and must be enclosed in square brackets. .if n .RE .RE .sp +\fBnlm\fP +.RS 4 +Get information about NLM (lockd) settings in the current net namespace. This +subcommand takes no arguments. +.RE +.sp \fBstatus\fP .RS 4 Get information about RPCs currently executing in the server. This subcommandutils/nfsdctl/nfsdctl.adoc +5
-0@@ -67,6 +67,11 @@ Each subcommand can also accept its own set of options and arguments. The addresses must be in dotted-quad form. IPv6 addresses should be in standard colon separated form, and must be enclosed in square brackets. +*nlm*:: + + Get information about NLM (lockd) settings in the current net namespace. This + subcommand takes no arguments. + *status*:: Get information about RPCs currently executing in the server. This subcommandutils/nfsdctl/nfsdctl.c +205
-13@@ -35,6 +35,12 @@#include "nfsd_netlink.h" #endif +#ifdef USE_SYSTEM_LOCKD_NETLINK_H +#include <linux/lockd_netlink.h> +#else +#include "lockd_netlink.h" +#endif +#include "nfsdctl.h" #include "conffile.h" #include "xlog.h" @@ -348,6 +354,28 @@ static void parse_pool_mode_get(struct genlmsghdr *gnlh) } } +static void parse_lockd_get (struct genlmsghdr *gnlh) +{ +struct nlattr *attr; +int rem; + +nla_for_each_attr (attr,genlmsg_attrdata (gnlh,0 ), +genlmsg_attrlen (gnlh,0 ), rem) { +switch (nla_type (attr)) { +case LOCKD_A_SERVER_GRACETIME : +printf ("gracetime: %u\n" ,nla_get_u32 (attr)); +break ; +case LOCKD_A_SERVER_TCP_PORT : +printf ("tcp_port: %hu\n" ,nla_get_u16 (attr)); +break ; +case LOCKD_A_SERVER_UDP_PORT : +printf ("udp_port: %hu\n" ,nla_get_u16 (attr)); +break ; +default : +break ; + } + } +}static int recv_handler (struct nl_msg *msg,void *arg) {struct genlmsghdr *gnlh =nlmsg_data (nlmsg_hdr (msg));@@ -368,6 +396,9 @@ static int recv_handler(struct nl_msg *msg, void *arg)case NFSD_CMD_POOL_MODE_GET :parse_pool_mode_get (gnlh);break ; +case LOCKD_CMD_SERVER_GET : +parse_lockd_get (gnlh); +break ;default :break ; }@@ -398,12 +429,12 @@ static struct nl_sock *netlink_sock_alloc(void)return sock; } -static struct nl_msg *netlink_msg_alloc (struct nl_sock *sock) +static struct nl_msg *netlink_msg_alloc (struct nl_sock *sock,const char *family) {struct nl_msg *msg;int id; - id =genl_ctrl_resolve (sock,NFSD_FAMILY_NAME ); + id =genl_ctrl_resolve (sock, family);if (id <0 ) {xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME );return NULL ;@@ -447,7 +478,7 @@ static int status_func(struct nl_sock *sock, int argc, char ** argv) } } - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -495,7 +526,7 @@ static int threads_doit(struct nl_sock *sock, int cmd, int grace, int lease,struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -607,7 +638,7 @@ static int fetch_nfsd_versions(struct nl_sock *sock)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -672,7 +703,7 @@ static int set_nfsd_versions(struct nl_sock *sock)struct nl_cb *cb;int i, ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -825,7 +856,7 @@ static int fetch_current_listeners(struct nl_sock *sock)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1054,7 +1085,7 @@ static int set_listeners(struct nl_sock *sock)struct nl_cb *cb;int i, ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1170,7 +1201,7 @@ static int pool_mode_doit(struct nl_sock *sock, int cmd, const char *pool_mode)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1249,6 +1280,131 @@ static int pool_mode_func(struct nl_sock *sock, int argc, char **argv)return pool_mode_doit (sock, cmd, pool_mode); } +static int lockd_config_doit (struct nl_sock *sock,int cmd,int grace,int tcpport,int udpport) +{ +struct genlmsghdr *ghdr; +struct nlmsghdr *nlh; +struct nl_msg *msg; +struct nl_cb *cb; +int ret; + +if (cmd ==LOCKD_CMD_SERVER_SET ) { +/* Do nothing if there is nothing to set */ +if (!grace && !tcpport && !udpport) +return 0 ; + } + + msg =netlink_msg_alloc (sock,LOCKD_FAMILY_NAME ); +if (!msg) +return 1 ; + + nlh =nlmsg_hdr (msg); +if (cmd ==LOCKD_CMD_SERVER_SET ) { +if (grace) +nla_put_u32 (msg,LOCKD_A_SERVER_GRACETIME , grace); +if (tcpport) +nla_put_u16 (msg,LOCKD_A_SERVER_TCP_PORT , tcpport); +if (udpport) +nla_put_u16 (msg,LOCKD_A_SERVER_UDP_PORT , udpport); + } + + ghdr =nlmsg_data (nlh); + ghdr->cmd = cmd; + + cb =nl_cb_alloc (NL_CB_CUSTOM ); +if (!cb) { +xlog (L_ERROR ,"failed to allocate netlink callbacks\n" ); + ret =1 ; + goto out; + } + + ret =nl_send_auto (sock, msg); +if (ret <0 ) { +xlog (L_ERROR ,"send failed (%d)!\n" , ret); + goto out_cb; + } + + ret =1 ; +nl_cb_err (cb,NL_CB_CUSTOM , error_handler, &ret); +nl_cb_set (cb,NL_CB_FINISH ,NL_CB_CUSTOM , finish_handler, &ret); +nl_cb_set (cb,NL_CB_ACK ,NL_CB_CUSTOM , ack_handler, &ret); +nl_cb_set (cb,NL_CB_VALID ,NL_CB_CUSTOM , recv_handler,NULL ); + +while (ret >0 ) +nl_recvmsgs (sock, cb); +if (ret <0 ) { +xlog (L_ERROR ,"Error: %s\n" ,strerror (-ret)); + ret =1 ; + } +out_cb: +nl_cb_put (cb); +out: +nlmsg_free (msg); +return ret; +} + +static int get_service (const char *svc) +{ +struct addrinfo *res, hints = { .ai_flags =AI_PASSIVE }; +int ret, port; + +if (!svc) +return 0 ; + + ret =getaddrinfo (NULL , svc, &hints, &res); +if (ret) { +xlog (L_ERROR ,"getaddrinfo of \"%s\" failed: %s\n" , + svc,gai_strerror (ret)); +return -1 ; + } + +switch (res->ai_family) { +case AF_INET : + { +struct sockaddr_in *sin = (struct sockaddr_in *)res->ai_addr; + + port =ntohs (sin->sin_port); + } +break ; +case AF_INET6 : + { +struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)res->ai_addr; + + port =ntohs (sin6->sin6_port); + } +break ; +default : +xlog (L_ERROR ,"Bad address family: %d\n" , res->ai_family); + port =-1 ; + } +freeaddrinfo (res); +return port; +} + +static int lockd_configure (struct nl_sock *sock,int grace) +{ +char *tcp_svc, *udp_svc; +int tcpport =0 , udpport =0 ; +int ret; + + tcp_svc =conf_get_str ("lockd" ,"port" ); +if (tcp_svc) { + tcpport =get_service (tcp_svc); +if (tcpport <0 ) +return 1 ; + } + + udp_svc =conf_get_str ("lockd" ,"udp-port" ); +if (udp_svc) { + udpport =get_service (udp_svc); +if (udpport <0 ) +return 1 ; + } + +return lockd_config_doit (sock,LOCKD_CMD_SERVER_SET , grace, tcpport, udpport); +} + +#define MAX_LISTENER_LEN (64 * 2 + 16) static void@@ -1355,6 +1511,13 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv)read_nfsd_conf (); + grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); + ret =lockd_configure (sock, grace); +if (ret) { +xlog (L_ERROR ,"lockd configuration failure" ); +return ret; + } + pool_mode =conf_get_str ("nfsd" ,"pool-mode" );if (pool_mode) { ret =pool_mode_doit (sock,NFSD_CMD_POOL_MODE_SET , pool_mode);@@ -1370,15 +1533,12 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv)if (ret)return ret; +xlog (D_GENERAL ,"configuring listeners" );configure_listeners (); ret =set_listeners (sock);if (ret)return ret; - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); - lease =conf_get_num ("nfsd" ,"lease-time" ,0 ); - scope =conf_get_str ("nfsd" ,"scope" ); - thread_str =conf_get_list ("nfsd" ,"threads" ); pools = thread_str ? thread_str->cnt :1 ;@@ -1402,6 +1562,9 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) threads[0 ] =DEFAULT_AUTOSTART_THREADS ; } + lease =conf_get_num ("nfsd" ,"lease-time" ,0 ); + scope =conf_get_str ("nfsd" ,"scope" ); + ret =threads_doit (sock,NFSD_CMD_THREADS_SET , grace, lease, pools, threads, scope); out:@@ -1409,6 +1572,30 @@ out:return ret; } +static void nlm_usage (void ) +{ +printf ("Usage: %s nlm\n" , taskname); +printf (" Get the current settings for the NLM (lockd) server.\n" ); +} + +static int nlm_func (struct nl_sock *sock,int argc,char ** argv) +{ +int *threads, grace, lease, idx, ret, opt, pools; +struct conf_list *thread_str; +struct conf_list_node *n; +char *scope, *pool_mode; + + optind =1 ; +while ((opt =getopt_long (argc, argv,"h" , help_only_options,NULL )) !=-1 ) { +switch (opt) { +case 'h' : +nlm_usage (); +return 0 ; + } + } +return lockd_config_doit (sock,LOCKD_CMD_SERVER_GET ,0 ,0 ,0 ); +} +enum nfsdctl_commands {NFSDCTL_STATUS ,NFSDCTL_THREADS ,@@ -1416,6 +1603,7 @@ enum nfsdctl_commands {NFSDCTL_LISTENER ,NFSDCTL_AUTOSTART ,NFSDCTL_POOL_MODE , + NFSDCTL_NLM, };static int parse_command (char *str)@@ -1432,6 +1620,8 @@ static int parse_command(char *str)return NFSDCTL_AUTOSTART ;if (!strcmp (str,"pool-mode" ))return NFSDCTL_POOL_MODE ; +if (!strcmp (str,"nlm" )) +return NFSDCTL_NLM ;return -1 ; }@@ -1444,6 +1634,7 @@ static nfsdctl_func func[] = { [NFSDCTL_LISTENER] = listener_func, [NFSDCTL_AUTOSTART] = autostart_func, [NFSDCTL_POOL_MODE] = pool_mode_func, + [NFSDCTL_NLM] = nlm_func, };static void usage (void )@@ -1460,6 +1651,7 @@ static void usage(void)printf (" listener get/set listener info\n" );printf (" version get/set supported NFS versions\n" );printf (" threads get/set nfsd thread settings\n" ); +printf (" nlm get current nlm settings\n" );printf (" status get current RPC processing info\n" );printf (" autostart start server with settings from /etc/nfs.conf\n" ); }
Signature · 1 line
Jeff Layton <[email protected]>
Tom Talpey
reviewerRe: [PATCH v2 3/3] nfsdctl: add necessary bits to configure lockd
Quoted 28 lines · Jeff Layton
Quoted 20 lines · Tom Talpey
Quoted 10 lines · Jeff Layton
The legacy rpc.nfsd program would configure the
nlm_grace_periodto match the NFSv4 grace period when starting the server. This adds similar functionality to the autostart subcommand using the new lockd netlink configuration interfaces. In addition, if lockd's udp or tcp listener ports are specified, also configure them before starting the server.A "nlm" subcommand is also added that will display the current lockd config settings in the current net ns. In the future, we could add the ability to set these values using the "nlm" subcommand, but for now I didn't see a real use-case for that, so I left it out.
It seems unnatural that the
nlm_grace_periodis tied to the netns.It seems to me it's more dependent on the network and its likely failure modes, the backend storage/filesystem, and perhaps the scale of clients performing possibly-conflicting locks. Oh, and also perhaps the minor version, since 4.1+ have the
RECLAIM_COMPLETEtermination event.Quoted 4 lines
Food for thought, perhaps.
Tom.
Fair point. More food:
Quoted 7 lines
My guess is that
nlm_grace_periodhandling is likely horribly broken in multi-container scenarios anyway. If you have multiple nfs server containers with different grace period settings, then they will all be competing to set the globalnlm_grace_period.Most clients end up reclaiming quickly enough that we don't hit major problems here, but it would be good to make this more "airtight".
The grace period can has been kicked down the road since nineteen ninety ump so, sure, what the heck :)
I wonder how many sysadmins are actually changing it, and why. It's not truly a correctness thing, more of an availability setting, so the choice of value is pretty soft. As you say, if the clients all reclaim promptly, nobody's the wiser. Do we have any data on how often the setting actually gets used?
Tom.
Quoted 469 lines
Quoted 468 lines
Quoted 467 lines
Signed-off-by Jeff Layton <[email protected]>--- configure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++++ utils/nfsdctl/nfsdctl.8 | 6 ++ utils/nfsdctl/nfsdctl.adoc | 5 + utils/nfsdctl/nfsdctl.c | 218 +++++++++++++++++++++++++++++++++++++++--- 5 files changed, 249 insertions(+), 13 deletions(-)configure.ac +4
-0@@ -268,6 +268,10 @@ AC_ARG_ENABLE(nfsdctl, [[int foo = NFSD_CMD_POOL_MODE_GET;]])], [AC_DEFINE([USE_SYSTEM_NFSD_NETLINK_H], 1, ["Use system's linux/nfsd_netlink.h"])]) + AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[#include <linux/lockd_netlink.h>]], + [[int foo = LOCKD_CMD_SERVER_GET;]])], + [AC_DEFINE([USE_SYSTEM_LOCKD_NETLINK_H], 1, + ["Use system's linux/lockd_netlink.h"])]) fi AC_ARG_ENABLE(nfsv4server,utils/nfsdctl/lockd_netlink.h new file +29
-0@@ -0,0 +1,29 @@ +/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */ +/* Do not edit directly, auto-generated from: */ +/* Documentation/netlink/specs/lockd.yaml */ +/* YNL-GEN uapi header */ + +#ifndef _UAPI_LINUX_LOCKD_NETLINK_H +#define _UAPI_LINUX_LOCKD_NETLINK_H + +#define LOCKD_FAMILY_NAME "lockd" +#define LOCKD_FAMILY_VERSION 1 + +enum { +LOCKD_A_SERVER_GRACETIME =1 , +LOCKD_A_SERVER_TCP_PORT , +LOCKD_A_SERVER_UDP_PORT , + + __LOCKD_A_SERVER_MAX, +LOCKD_A_SERVER_MAX = (__LOCKD_A_SERVER_MAX -1 ) +}; + +enum { +LOCKD_CMD_SERVER_SET =1 , +LOCKD_CMD_SERVER_GET , + + __LOCKD_CMD_MAX, +LOCKD_CMD_MAX = (__LOCKD_CMD_MAX -1 ) +}; + +#endif /* _UAPI_LINUX_LOCKD_NETLINK_H */ utils/nfsdctl/nfsdctl.8 +6
-0@@ -127,6 +127,12 @@ colon separated form, and must be enclosed in square brackets. .if n .RE .RE .sp +\fBnlm\fP +.RS 4 +Get information about NLM (lockd) settings in the current net namespace. This +subcommand takes no arguments. +.RE +.sp \fBstatus\fP .RS 4 Get information about RPCs currently executing in the server. This subcommandutils/nfsdctl/nfsdctl.adoc +5
-0@@ -67,6 +67,11 @@ Each subcommand can also accept its own set of options and arguments. The addresses must be in dotted-quad form. IPv6 addresses should be in standard colon separated form, and must be enclosed in square brackets. +*nlm*:: + + Get information about NLM (lockd) settings in the current net namespace. This + subcommand takes no arguments. + *status*:: Get information about RPCs currently executing in the server. This subcommandutils/nfsdctl/nfsdctl.c +205
-13@@ -35,6 +35,12 @@#include "nfsd_netlink.h" #endif +#ifdef USE_SYSTEM_LOCKD_NETLINK_H +#include <linux/lockd_netlink.h> +#else +#include "lockd_netlink.h" +#endif +#include "nfsdctl.h" #include "conffile.h" #include "xlog.h" @@ -348,6 +354,28 @@ static void parse_pool_mode_get(struct genlmsghdr *gnlh) } } +static void parse_lockd_get (struct genlmsghdr *gnlh) +{ +struct nlattr *attr; +int rem; + +nla_for_each_attr (attr,genlmsg_attrdata (gnlh,0 ), +genlmsg_attrlen (gnlh,0 ), rem) { +switch (nla_type (attr)) { +case LOCKD_A_SERVER_GRACETIME : +printf ("gracetime: %u\n" ,nla_get_u32 (attr)); +break ; +case LOCKD_A_SERVER_TCP_PORT : +printf ("tcp_port: %hu\n" ,nla_get_u16 (attr)); +break ; +case LOCKD_A_SERVER_UDP_PORT : +printf ("udp_port: %hu\n" ,nla_get_u16 (attr)); +break ; +default : +break ; + } + } +}static int recv_handler (struct nl_msg *msg,void *arg) {struct genlmsghdr *gnlh =nlmsg_data (nlmsg_hdr (msg));@@ -368,6 +396,9 @@ static int recv_handler(struct nl_msg *msg, void *arg)case NFSD_CMD_POOL_MODE_GET :parse_pool_mode_get (gnlh);break ; +case LOCKD_CMD_SERVER_GET : +parse_lockd_get (gnlh); +break ;default :break ; }@@ -398,12 +429,12 @@ static struct nl_sock *netlink_sock_alloc(void)return sock; } -static struct nl_msg *netlink_msg_alloc (struct nl_sock *sock) +static struct nl_msg *netlink_msg_alloc (struct nl_sock *sock,const char *family) {struct nl_msg *msg;int id; - id =genl_ctrl_resolve (sock,NFSD_FAMILY_NAME ); + id =genl_ctrl_resolve (sock, family);if (id <0 ) {xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME );return NULL ;@@ -447,7 +478,7 @@ static int status_func(struct nl_sock *sock, int argc, char ** argv) } } - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -495,7 +526,7 @@ static int threads_doit(struct nl_sock *sock, int cmd, int grace, int lease,struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -607,7 +638,7 @@ static int fetch_nfsd_versions(struct nl_sock *sock)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -672,7 +703,7 @@ static int set_nfsd_versions(struct nl_sock *sock)struct nl_cb *cb;int i, ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -825,7 +856,7 @@ static int fetch_current_listeners(struct nl_sock *sock)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1054,7 +1085,7 @@ static int set_listeners(struct nl_sock *sock)struct nl_cb *cb;int i, ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1170,7 +1201,7 @@ static int pool_mode_doit(struct nl_sock *sock, int cmd, const char *pool_mode)struct nl_cb *cb;int ret; - msg =netlink_msg_alloc (sock); + msg =netlink_msg_alloc (sock,NFSD_FAMILY_NAME );if (!msg)return 1 ;@@ -1249,6 +1280,131 @@ static int pool_mode_func(struct nl_sock *sock, int argc, char **argv)return pool_mode_doit (sock, cmd, pool_mode); } +static int lockd_config_doit (struct nl_sock *sock,int cmd,int grace,int tcpport,int udpport) +{ +struct genlmsghdr *ghdr; +struct nlmsghdr *nlh; +struct nl_msg *msg; +struct nl_cb *cb; +int ret; + +if (cmd ==LOCKD_CMD_SERVER_SET ) { +/* Do nothing if there is nothing to set */ +if (!grace && !tcpport && !udpport) +return 0 ; + } + + msg =netlink_msg_alloc (sock,LOCKD_FAMILY_NAME ); +if (!msg) +return 1 ; + + nlh =nlmsg_hdr (msg); +if (cmd ==LOCKD_CMD_SERVER_SET ) { +if (grace) +nla_put_u32 (msg,LOCKD_A_SERVER_GRACETIME , grace); +if (tcpport) +nla_put_u16 (msg,LOCKD_A_SERVER_TCP_PORT , tcpport); +if (udpport) +nla_put_u16 (msg,LOCKD_A_SERVER_UDP_PORT , udpport); + } + + ghdr =nlmsg_data (nlh); + ghdr->cmd = cmd; + + cb =nl_cb_alloc (NL_CB_CUSTOM ); +if (!cb) { +xlog (L_ERROR ,"failed to allocate netlink callbacks\n" ); + ret =1 ; + goto out; + } + + ret =nl_send_auto (sock, msg); +if (ret <0 ) { +xlog (L_ERROR ,"send failed (%d)!\n" , ret); + goto out_cb; + } + + ret =1 ; +nl_cb_err (cb,NL_CB_CUSTOM , error_handler, &ret); +nl_cb_set (cb,NL_CB_FINISH ,NL_CB_CUSTOM , finish_handler, &ret); +nl_cb_set (cb,NL_CB_ACK ,NL_CB_CUSTOM , ack_handler, &ret); +nl_cb_set (cb,NL_CB_VALID ,NL_CB_CUSTOM , recv_handler,NULL ); + +while (ret >0 ) +nl_recvmsgs (sock, cb); +if (ret <0 ) { +xlog (L_ERROR ,"Error: %s\n" ,strerror (-ret)); + ret =1 ; + } +out_cb: +nl_cb_put (cb); +out: +nlmsg_free (msg); +return ret; +} + +static int get_service (const char *svc) +{ +struct addrinfo *res, hints = { .ai_flags =AI_PASSIVE }; +int ret, port; + +if (!svc) +return 0 ; + + ret =getaddrinfo (NULL , svc, &hints, &res); +if (ret) { +xlog (L_ERROR ,"getaddrinfo of \"%s\" failed: %s\n" , + svc,gai_strerror (ret)); +return -1 ; + } + +switch (res->ai_family) { +case AF_INET : + { +struct sockaddr_in *sin = (struct sockaddr_in *)res->ai_addr; + + port =ntohs (sin->sin_port); + } +break ; +case AF_INET6 : + { +struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)res->ai_addr; + + port =ntohs (sin6->sin6_port); + } +break ; +default : +xlog (L_ERROR ,"Bad address family: %d\n" , res->ai_family); + port =-1 ; + } +freeaddrinfo (res); +return port; +} + +static int lockd_configure (struct nl_sock *sock,int grace) +{ +char *tcp_svc, *udp_svc; +int tcpport =0 , udpport =0 ; +int ret; + + tcp_svc =conf_get_str ("lockd" ,"port" ); +if (tcp_svc) { + tcpport =get_service (tcp_svc); +if (tcpport <0 ) +return 1 ; + } + + udp_svc =conf_get_str ("lockd" ,"udp-port" ); +if (udp_svc) { + udpport =get_service (udp_svc); +if (udpport <0 ) +return 1 ; + } + +return lockd_config_doit (sock,LOCKD_CMD_SERVER_SET , grace, tcpport, udpport); +} + +#define MAX_LISTENER_LEN (64 * 2 + 16) static void@@ -1355,6 +1511,13 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv)read_nfsd_conf (); + grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); + ret =lockd_configure (sock, grace); +if (ret) { +xlog (L_ERROR ,"lockd configuration failure" ); +return ret; + } + pool_mode =conf_get_str ("nfsd" ,"pool-mode" );if (pool_mode) { ret =pool_mode_doit (sock,NFSD_CMD_POOL_MODE_SET , pool_mode);@@ -1370,15 +1533,12 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv)if (ret)return ret; +xlog (D_GENERAL ,"configuring listeners" );configure_listeners (); ret =set_listeners (sock);if (ret)return ret; - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); - lease =conf_get_num ("nfsd" ,"lease-time" ,0 ); - scope =conf_get_str ("nfsd" ,"scope" ); - thread_str =conf_get_list ("nfsd" ,"threads" ); pools = thread_str ? thread_str->cnt :1 ;@@ -1402,6 +1562,9 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) threads[0 ] =DEFAULT_AUTOSTART_THREADS ; } + lease =conf_get_num ("nfsd" ,"lease-time" ,0 ); + scope =conf_get_str ("nfsd" ,"scope" ); + ret =threads_doit (sock,NFSD_CMD_THREADS_SET , grace, lease, pools, threads, scope); out:@@ -1409,6 +1572,30 @@ out:return ret; } +static void nlm_usage (void ) +{ +printf ("Usage: %s nlm\n" , taskname); +printf (" Get the current settings for the NLM (lockd) server.\n" ); +} + +static int nlm_func (struct nl_sock *sock,int argc,char ** argv) +{ +int *threads, grace, lease, idx, ret, opt, pools; +struct conf_list *thread_str; +struct conf_list_node *n; +char *scope, *pool_mode; + + optind =1 ; +while ((opt =getopt_long (argc, argv,"h" , help_only_options,NULL )) !=-1 ) { +switch (opt) { +case 'h' : +nlm_usage (); +return 0 ; + } + } +return lockd_config_doit (sock,LOCKD_CMD_SERVER_GET ,0 ,0 ,0 ); +} +enum nfsdctl_commands {NFSDCTL_STATUS ,NFSDCTL_THREADS ,@@ -1416,6 +1603,7 @@ enum nfsdctl_commands {NFSDCTL_LISTENER ,NFSDCTL_AUTOSTART ,NFSDCTL_POOL_MODE , + NFSDCTL_NLM, };static int parse_command (char *str)@@ -1432,6 +1620,8 @@ static int parse_command(char *str)return NFSDCTL_AUTOSTART ;if (!strcmp (str,"pool-mode" ))return NFSDCTL_POOL_MODE ; +if (!strcmp (str,"nlm" )) +return NFSDCTL_NLM ;return -1 ; }@@ -1444,6 +1634,7 @@ static nfsdctl_func func[] = { [NFSDCTL_LISTENER] = listener_func, [NFSDCTL_AUTOSTART] = autostart_func, [NFSDCTL_POOL_MODE] = pool_mode_func, + [NFSDCTL_NLM] = nlm_func, };static void usage (void )@@ -1460,6 +1651,7 @@ static void usage(void)printf (" listener get/set listener info\n" );printf (" version get/set supported NFS versions\n" );printf (" threads get/set nfsd thread settings\n" ); +printf (" nlm get current nlm settings\n" );printf (" status get current RPC processing info\n" );printf (" autostart start server with settings from /etc/nfs.conf\n" ); }
Re: [PATCH v2 3/3] nfsdctl: add necessary bits to configure lockd
Quoted 5 lines · Tom Talpey
I wonder how many sysadmins are actually changing it, and why. It's not truly a correctness thing, more of an availability setting, so the choice of value is pretty soft. As you say, if the clients all reclaim promptly, nobody's the wiser. Do we have any data on how often the setting actually gets used?
Here's an anecdote - when I was a sysadmin (~12 years ago), we were very aware of the grace period and tried to optimize it because too long meant many of our clients applications would hang for a longer outage, and too short meant that all the locks weren't recovered.
We also depended on lock recovery working well because we would often update and migrate servers.
This was before knfsd v4 could end it early, which is.. quite nice.
Ben
Tom Talpey
reviewerRe: [PATCH v2 3/3] nfsdctl: add necessary bits to configure lockd
Quoted 8 lines · Benjamin Coddington
Quoted 5 lines · Tom Talpey
I wonder how many sysadmins are actually changing it, and why. It's not truly a correctness thing, more of an availability setting, so the choice of value is pretty soft. As you say, if the clients all reclaim promptly, nobody's the wiser. Do we have any data on how often the setting actually gets used?
Quoted 7 lines
Here's an anecdote - when I was a sysadmin (~12 years ago), we were very aware of the grace period and tried to optimize it because too long meant many of our clients applications would hang for a longer outage, and too short meant that all the locks weren't recovered.
We also depended on lock recovery working well because we would often update and migrate servers.
Ok good to know - v3 and v4.0 I assume.
Quoted 1 line
This was before knfsd v4 could end it early, which is.. quite nice.
Yep, v4.1 RECLAIM_COMPLETE, right? You're welcome. :)
Tom.
Quoted 7 lines · Jeff Layton
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698
Quoted 8 lines
There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
Signed-off-by Jeff Layton <[email protected]>
I think the read_nfsd_conf call should be moved out of autostart_func and into main (right before the command-line options are parsed). Right now if you enable debugging in nfs.conf, you get the "configuring listeners" and "nfsdctl exiting" messages, but not the "nfsdctl started" message. It's not a big deal though and could be done if additional debug logging is added in the future.
Quoted 27 lines
---
Changes in v2:
- properly regenerate manpages
- fix up bogus merge conflict
- addD_GENERALxlog messages when nfsdctl starts and exits
- Link to v1: https://lore.kernel.org/r/[email protected]---
Jeff Layton (3):
nfsdctl: convert toxlog()
nfsdctl: fix the --version option
nfsdctl: add necessary bits to configure lockdconfigure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++ utils/nfsdctl/nfsdctl.8 | 15 +- utils/nfsdctl/nfsdctl.adoc | 8 + utils/nfsdctl/nfsdctl.c | 331 ++++++++++++++++++++++++++++++++++-------- 5 files changed, 321 insertions(+), 66 deletions(-)---
base-commit:65f4cc3a6ce1472ee4092c4bbf4b19beb0a8217b
change-id: 20250109-lockd-nl-6272fa9e8a5dBest regards,
Signature · 1 line
Jeff Layton <[email protected]>
Jeff Layton
maintainerQuoted 16 lines · Scott Mayhew
Quoted 14 lines · Jeff Layton
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
Quoted 9 lines
Quoted 1 line
Signed-off-by Jeff Layton <[email protected]>I think the
read_nfsd_confcall should be moved out ofautostart_funcand into main (right before the command-line options are parsed). Right now if you enable debugging in nfs.conf, you get the "configuring listeners" and "nfsdctl exiting" messages, but not the "nfsdctl started" message. It's not a big deal though and could be done if additional debug logging is added in the future.
That sounds good. We can do that in a separate patch.
Quoted 1 line
Reviewed-by Scott Mayhew <[email protected]>
Thanks!
Quoted 28 lines
Quoted 27 lines
---
Changes in v2:
- properly regenerate manpages
- fix up bogus merge conflict
- addD_GENERALxlog messages when nfsdctl starts and exits
- Link to v1: https://lore.kernel.org/r/[email protected]---
Jeff Layton (3):
nfsdctl: convert toxlog()
nfsdctl: fix the --version option
nfsdctl: add necessary bits to configure lockdconfigure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++ utils/nfsdctl/nfsdctl.8 | 15 +- utils/nfsdctl/nfsdctl.adoc | 8 + utils/nfsdctl/nfsdctl.c | 331 ++++++++++++++++++++++++++++++++++-------- 5 files changed, 321 insertions(+), 66 deletions(-)---
base-commit:65f4cc3a6ce1472ee4092c4bbf4b19beb0a8217b
change-id: 20250109-lockd-nl-6272fa9e8a5dBest regards,
Signature · 1 line
Jeff Layton <[email protected]>
Signature · 1 line
Jeff Layton <[email protected]>
Quoted 27 lines · Jeff Layton
Quoted 16 lines · Scott Mayhew
Quoted 14 lines · Jeff Layton
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
Quoted 9 lines
Quoted 1 line
Signed-off-by Jeff Layton <[email protected]>I think the
read_nfsd_confcall should be moved out ofautostart_funcand into main (right before the command-line options are parsed). Right now if you enable debugging in nfs.conf, you get the "configuring listeners" and "nfsdctl exiting" messages, but not the "nfsdctl started" message. It's not a big deal though and could be done if additional debug logging is added in the future.
Quoted 5 lines
That sounds good. We can do that in a separate patch.
Quoted 1 line
Reviewed-by Scott Mayhew <[email protected]>Thanks!
Hey, Jeff. I was testing this against a kernel without the lockd netlink patch, and I get this:
Jan 15 09:39:16 systemd[1]: Starting nfs-server.service - NFS server and services... Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl started Jan 15 09:39:17 sh[1603]: nfsdctl: nfsd not found Jan 15 09:39:17 sh[1603]: nfsdctl: lockd configuration failure Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl exiting Jan 15 09:39:17 sh[1601]: rpc.nfsd: knfsd is currently down Jan 15 09:39:17 sh[1601]: rpc.nfsd: Writing version string to kernel: -2 +3 +4 +4.1 +4.2 Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET TCP socket. Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET6 TCP socket.
Do we really want it falling back to rpc.nfsd if it can't configure lockd? Maybe it should emit a warning instead?
At the very least, NFSD_FAMILY_NAME should no longer be hard-coded in that "not found" error message in netlink_msg_alloc().
-Scott
Quoted 33 lines
Quoted 29 lines
Quoted 28 lines
---
Changes in v2:
- properly regenerate manpages
- fix up bogus merge conflict
- addD_GENERALxlog messages when nfsdctl starts and exits
- Link to v1: https://lore.kernel.org/r/[email protected]---
Jeff Layton (3):
nfsdctl: convert toxlog()
nfsdctl: fix the --version option
nfsdctl: add necessary bits to configure lockdconfigure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++ utils/nfsdctl/nfsdctl.8 | 15 +- utils/nfsdctl/nfsdctl.adoc | 8 + utils/nfsdctl/nfsdctl.c | 331 ++++++++++++++++++++++++++++++++++-------- 5 files changed, 321 insertions(+), 66 deletions(-)---
base-commit:65f4cc3a6ce1472ee4092c4bbf4b19beb0a8217b
change-id: 20250109-lockd-nl-6272fa9e8a5dBest regards,
Signature · 1 line
Jeff Layton <[email protected]>Signature · 1 line
Jeff Layton <[email protected]>
Jeff Layton
maintainerQuoted 48 lines · Scott Mayhew
Quoted 27 lines · Jeff Layton
Quoted 16 lines · Scott Mayhew
Quoted 14 lines · Jeff Layton
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
Quoted 9 lines
Quoted 1 line
Signed-off-by Jeff Layton <[email protected]>I think the
read_nfsd_confcall should be moved out ofautostart_funcand into main (right before the command-line options are parsed). Right now if you enable debugging in nfs.conf, you get the "configuring listeners" and "nfsdctl exiting" messages, but not the "nfsdctl started" message. It's not a big deal though and could be done if additional debug logging is added in the future.Quoted 5 lines
That sounds good. We can do that in a separate patch.
Quoted 1 line
Reviewed-by Scott Mayhew <[email protected]>Thanks!
Hey, Jeff. I was testing this against a kernel without the lockd netlink patch, and I get this:
Jan 15 09:39:16 systemd[1]: Starting nfs-server.service - NFS server and services... Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl started Jan 15 09:39:17 sh[1603]: nfsdctl: nfsd not found Jan 15 09:39:17 sh[1603]: nfsdctl: lockd configuration failure Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl exiting Jan 15 09:39:17 sh[1601]: rpc.nfsd: knfsd is currently down Jan 15 09:39:17 sh[1601]: rpc.nfsd: Writing version string to kernel: -2 +3 +4 +4.1 +4.2 Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET TCP socket. Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET6 TCP socket.
Quoted 3 lines
Do we really want it falling back to rpc.nfsd if it can't configure lockd? Maybe it should emit a warning instead?
I thought about that, and I think it's better to error out here.
Falling back to rpc.nfsd is harmless, and only people who are trying to set the grace period or lockd ports will ever hit this. lockd configuration is a no-op if none of those settings are set.
Quoted 3 lines
At the very least,
NFSD_FAMILY_NAMEshould no longer be hard-coded in that "not found" error message innetlink_msg_alloc().
Yeah, that would be good to fix.
Quoted 35 lines
Quoted 34 lines
Quoted 30 lines
Quoted 29 lines
---
Changes in v2:
- properly regenerate manpages
- fix up bogus merge conflict
- addD_GENERALxlog messages when nfsdctl starts and exits
- Link to v1: https://lore.kernel.org/r/[email protected]---
Jeff Layton (3):
nfsdctl: convert toxlog()
nfsdctl: fix the --version option
nfsdctl: add necessary bits to configure lockdconfigure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++ utils/nfsdctl/nfsdctl.8 | 15 +- utils/nfsdctl/nfsdctl.adoc | 8 + utils/nfsdctl/nfsdctl.c | 331 ++++++++++++++++++++++++++++++++++-------- 5 files changed, 321 insertions(+), 66 deletions(-)---
base-commit:65f4cc3a6ce1472ee4092c4bbf4b19beb0a8217b
change-id: 20250109-lockd-nl-6272fa9e8a5dBest regards,
Signature · 1 line
Jeff Layton <[email protected]>Signature · 1 line
Jeff Layton <[email protected]>
Signature · 1 line
Jeff Layton <[email protected]>
Quoted 59 lines · Jeff Layton
Quoted 48 lines · Scott Mayhew
Quoted 27 lines · Jeff Layton
Quoted 16 lines · Scott Mayhew
Quoted 14 lines · Jeff Layton
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
Quoted 9 lines
Quoted 1 line
Signed-off-by Jeff Layton <[email protected]>I think the
read_nfsd_confcall should be moved out ofautostart_funcand into main (right before the command-line options are parsed). Right now if you enable debugging in nfs.conf, you get the "configuring listeners" and "nfsdctl exiting" messages, but not the "nfsdctl started" message. It's not a big deal though and could be done if additional debug logging is added in the future.Quoted 5 lines
That sounds good. We can do that in a separate patch.
Quoted 1 line
Reviewed-by Scott Mayhew <[email protected]>Thanks!
Hey, Jeff. I was testing this against a kernel without the lockd netlink patch, and I get this:
Jan 15 09:39:16 systemd[1]: Starting nfs-server.service - NFS server and services... Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl started Jan 15 09:39:17 sh[1603]: nfsdctl: nfsd not found Jan 15 09:39:17 sh[1603]: nfsdctl: lockd configuration failure Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl exiting Jan 15 09:39:17 sh[1601]: rpc.nfsd: knfsd is currently down Jan 15 09:39:17 sh[1601]: rpc.nfsd: Writing version string to kernel: -2 +3 +4 +4.1 +4.2 Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET TCP socket. Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET6 TCP socket.Quoted 3 lines
Do we really want it falling back to rpc.nfsd if it can't configure lockd? Maybe it should emit a warning instead?
I thought about that, and I think it's better to error out here.
Falling back to rpc.nfsd is harmless, and only people who are trying to set the grace period or lockd ports will ever hit this. lockd configuration is a no-op if none of those settings are set.
Quoted 6 lines
Quoted 3 lines
At the very least,
NFSD_FAMILY_NAMEshould no longer be hard-coded in that "not found" error message innetlink_msg_alloc().Yeah, that would be good to fix.
On a rawhide kernel (6.13.0-0.rc6) the server does come up with 'nfsdctl autostart' but with the new argument 'nlm' I'm getting
$ nfsdctl nlm nfsdctl: nfsd not found
steved.
Jeff Layton
maintainerQuoted 68 lines · Steve Dickson
Quoted 59 lines · Jeff Layton
Quoted 48 lines · Scott Mayhew
Quoted 27 lines · Jeff Layton
Quoted 16 lines · Scott Mayhew
Quoted 14 lines · Jeff Layton
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
Quoted 9 lines
Quoted 1 line
Signed-off-by Jeff Layton <[email protected]>I think the
read_nfsd_confcall should be moved out ofautostart_funcand into main (right before the command-line options are parsed). Right now if you enable debugging in nfs.conf, you get the "configuring listeners" and "nfsdctl exiting" messages, but not the "nfsdctl started" message. It's not a big deal though and could be done if additional debug logging is added in the future.Quoted 5 lines
That sounds good. We can do that in a separate patch.
Quoted 1 line
Reviewed-by Scott Mayhew <[email protected]>Thanks!
Hey, Jeff. I was testing this against a kernel without the lockd netlink patch, and I get this:
Jan 15 09:39:16 systemd[1]: Starting nfs-server.service - NFS server and services... Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl started Jan 15 09:39:17 sh[1603]: nfsdctl: nfsd not found Jan 15 09:39:17 sh[1603]: nfsdctl: lockd configuration failure Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl exiting Jan 15 09:39:17 sh[1601]: rpc.nfsd: knfsd is currently down Jan 15 09:39:17 sh[1601]: rpc.nfsd: Writing version string to kernel: -2 +3 +4 +4.1 +4.2 Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET TCP socket. Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET6 TCP socket.Quoted 3 lines
Do we really want it falling back to rpc.nfsd if it can't configure lockd? Maybe it should emit a warning instead?
I thought about that, and I think it's better to error out here.
Falling back to rpc.nfsd is harmless, and only people who are trying to set the grace period or lockd ports will ever hit this. lockd configuration is a no-op if none of those settings are set.
Quoted 6 lines
Quoted 3 lines
At the very least,
NFSD_FAMILY_NAMEshould no longer be hard-coded in that "not found" error message innetlink_msg_alloc().Yeah, that would be good to fix.
Quoted 7 lines
On a rawhide kernel (6.13.0-0.rc6) the server does come up with 'nfsdctl autostart' but with the new argument 'nlm' I'm getting
$ nfsdctl nlm nfsdctl: nfsd not found
Yeah, that's what Scott pointed out too. We should make that error message a bit more friendly. It may be a bit before I can get to it. Do you guys want to propose a patch to fix that?
Thanks,
Signature · 1 line
Jeff Layton <[email protected]>
Quoted 73 lines · Jeff Layton
Quoted 72 lines · Steve Dickson
Quoted 59 lines · Jeff Layton
Quoted 48 lines · Scott Mayhew
Quoted 27 lines · Jeff Layton
Quoted 16 lines · Scott Mayhew
Quoted 14 lines · Jeff Layton
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
Quoted 9 lines
Quoted 1 line
Signed-off-by Jeff Layton <[email protected]>I think the
read_nfsd_confcall should be moved out ofautostart_funcand into main (right before the command-line options are parsed). Right now if you enable debugging in nfs.conf, you get the "configuring listeners" and "nfsdctl exiting" messages, but not the "nfsdctl started" message. It's not a big deal though and could be done if additional debug logging is added in the future.Quoted 5 lines
That sounds good. We can do that in a separate patch.
Quoted 1 line
Reviewed-by Scott Mayhew <[email protected]>Thanks!
Hey, Jeff. I was testing this against a kernel without the lockd netlink patch, and I get this:
Jan 15 09:39:16 systemd[1]: Starting nfs-server.service - NFS server and services... Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl started Jan 15 09:39:17 sh[1603]: nfsdctl: nfsd not found Jan 15 09:39:17 sh[1603]: nfsdctl: lockd configuration failure Jan 15 09:39:17 sh[1603]: nfsdctl: nfsdctl exiting Jan 15 09:39:17 sh[1601]: rpc.nfsd: knfsd is currently down Jan 15 09:39:17 sh[1601]: rpc.nfsd: Writing version string to kernel: -2 +3 +4 +4.1 +4.2 Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET TCP socket. Jan 15 09:39:17 sh[1601]: rpc.nfsd: Created AF_INET6 TCP socket.Quoted 3 lines
Do we really want it falling back to rpc.nfsd if it can't configure lockd? Maybe it should emit a warning instead?
I thought about that, and I think it's better to error out here.
Falling back to rpc.nfsd is harmless, and only people who are trying to set the grace period or lockd ports will ever hit this. lockd configuration is a no-op if none of those settings are set.
Quoted 6 lines
Quoted 3 lines
At the very least,
NFSD_FAMILY_NAMEshould no longer be hard-coded in that "not found" error message innetlink_msg_alloc().Yeah, that would be good to fix.
On a rawhide kernel (6.13.0-0.rc6) the server does come up with 'nfsdctl autostart' but with the new argument 'nlm' I'm getting
Quoted 7 lines
Quoted 3 lines
$ nfsdctl nlm nfsdctl: nfsd not foundYeah, that's what Scott pointed out too. We should make that error message a bit more friendly. It may be a bit before I can get to it. Do you guys want to propose a patch to fix that?
Sure, I can do that.
Quoted 5 lines
Thanks,
Signature · 1 line
Jeff Layton <[email protected]>
PATCH 1/1 nfsdctl: debug logging fixups
In reply to Jeff LaytonMove read_nfsd_conf() out of autostart_func() and into main(). Remove hard-coded NFSD_FAMILY_NAME in the first error message in netlink_msg_alloc() and make the error messages in netlink_msg_alloc() more descriptive/unique.
---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
utils/nfsdctl/nfsdctl.c +4 -4
@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }
@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }
@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {
@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {
Signature · 1 line
2.48.0
Re: [nfs-utils PATCH] nfsdctl: debug logging fixups
Quoted 45 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {
utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {
LGTM!
Re: [nfs-utils PATCH] nfsdctl: debug logging fixups
In reply to Scott MayhewQuoted 9 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.
Got it...
Quoted 36 lines
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {
utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {
Ok... at this point we a prettier error message
$ nfsdctl nlm nfsdctl: failed to resolve lockd generic netlink family
But the point of this argument is:
Get information about NLM (lockd) settings in the current net namespace. This subcommand takes no arguments.
How is that giving information from the running lockd?
What am I missing??
steved.
Jeff Layton
maintainerRe: [nfs-utils PATCH] nfsdctl: debug logging fixups
Quoted 61 lines · Steve Dickson
Quoted 9 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.Got it...
Quoted 36 lines
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {Ok... at this point we a prettier error message
$ nfsdctl nlm nfsdctl: failed to resolve lockd generic netlink family
Quoted 9 lines
But the point of this argument is:
Get information about NLM (lockd) settings in the current net namespace. This subcommand takes no arguments.
How is that giving information from the running lockd?
What am I missing??
You're missing a kernel that has the required netlink interface. To test this properly, you'll need to patch your kernel, until that patch makes it upstream.
Signature · 1 line
Jeff Layton <[email protected]>
Re: [nfs-utils PATCH] nfsdctl: debug logging fixups
Quoted 67 lines · Jeff Layton
Quoted 66 lines · Steve Dickson
Quoted 9 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.Got it...
Quoted 36 lines
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {Ok... at this point we a prettier error message
$ nfsdctl nlm nfsdctl: failed to resolve lockd generic netlink familyBut the point of this argument is:
Get information about NLM (lockd) settings in the current net namespace. This subcommand takes no arguments.
Quoted 8 lines
Quoted 4 lines
How is that giving information from the running lockd?
What am I missing??
You're missing a kernel that has the required netlink interface. To test this properly, you'll need to patch your kernel, until that patch makes it upstream.
Okay... I figured it was something like that. But doesn't make sense to wait until the patch is in upstream so the argument can be properly tested? Why add an argument that will always fail?
steved.
Jeff Layton
maintainerRe: [nfs-utils PATCH] nfsdctl: debug logging fixups
Quoted 74 lines · Steve Dickson
Quoted 72 lines · Jeff Layton
Quoted 70 lines · Steve Dickson
Quoted 9 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.Got it...
Quoted 36 lines
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {Ok... at this point we a prettier error message
$ nfsdctl nlm nfsdctl: failed to resolve lockd generic netlink familyBut the point of this argument is:
Get information about NLM (lockd) settings in the current net namespace. This subcommand takes no arguments.
How is that giving information from the running lockd?
What am I missing??
Quoted 7 lines
Quoted 3 lines
You're missing a kernel that has the required netlink interface. To test this properly, you'll need to patch your kernel, until that patch makes it upstream.
Okay... I figured it was something like that. But doesn't make sense to wait until the patch is in upstream so the argument can be properly tested? Why add an argument that will always fail?
Why can't it be properly tested? It's just a matter of running a more recent kernel that has the right interfaces. That should be in linux- next soon (if not already).
I think the question is whether we want to wait until the kernel interfaces trickle out into downstream distro kernels before we ship any userland support in an upstream project (nfs-utils).
If you want to wait until it hits Fedora Rawhide kernels, then you're looking at about 10-12 weeks from now. If you want to wait until it makes it into a stable Fedora release kernel then we're looking at about 6 months from now.
I'll note that that it took 6 months to get the original nfsdctl patches merged because of the lag on kernel patches making it into distros, and I think that was way too long.
Signature · 1 line
Jeff Layton <[email protected]>
Re: [nfs-utils PATCH] nfsdctl: debug logging fixups
Quoted 78 lines · Jeff Layton
Quoted 77 lines · Steve Dickson
Quoted 75 lines · Jeff Layton
Quoted 61 lines · Steve Dickson
Quoted 9 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.Got it...
Quoted 36 lines
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {Ok... at this point we a prettier error message
$ nfsdctl nlm nfsdctl: failed to resolve lockd generic netlink familyQuoted 9 lines
But the point of this argument is:
Get information about NLM (lockd) settings in the current net namespace. This subcommand takes no arguments.
How is that giving information from the running lockd?
What am I missing??
You're missing a kernel that has the required netlink interface. To test this properly, you'll need to patch your kernel, until that patch makes it upstream.
Quoted 8 lines
Quoted 4 lines
Okay... I figured it was something like that. But doesn't make sense to wait until the patch is in upstream so the argument can be properly tested? Why add an argument that will always fail?
Why can't it be properly tested? It's just a matter of running a more recent kernel that has the right interfaces. That should be in linux- next soon (if not already).
I'm doing my testing on a 6.13.0-0.rc6 which will soon be a 6.14 kernel... its my understanding the needed kernel patch will be in the 6.15 kernel... Please correct me if that is not true.
Quoted 4 lines
I think the question is whether we want to wait until the kernel interfaces trickle out into downstream distro kernels before we ship any userland support in an upstream project (nfs-utils).
Yes! As soon as the kernel support hits the upstream kernel, we will be good to go. I just don't want to put a feature in that will fail %100 of the time.
Quoted 5 lines
If you want to wait until it hits Fedora Rawhide kernels, then you're looking at about 10-12 weeks from now. If you want to wait until it makes it into a stable Fedora release kernel then we're looking at about 6 months from now.
nfsdctl is in all current Fedora stable releases, which is the reason I'm pushing back. I do not want to put something in that will make it fail. That just does not make sense to me.
Quoted 4 lines
I'll note that that it took 6 months to get the original nfsdctl patches merged because of the lag on kernel patches making it into distros, and I think that was way too long.
It took that long because there were issues with the command. In which I was glad to help debug some of the issues...
New technology takes time to develop... I just think this is one of those cases.
steved.
Jeff Layton
maintainerRe: [nfs-utils PATCH] nfsdctl: debug logging fixups
Quoted 111 lines · Steve Dickson
Quoted 78 lines · Jeff Layton
Quoted 77 lines · Steve Dickson
Quoted 75 lines · Jeff Layton
Quoted 61 lines · Steve Dickson
Quoted 9 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.Got it...
Quoted 36 lines
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {Ok... at this point we a prettier error message
$ nfsdctl nlm nfsdctl: failed to resolve lockd generic netlink familyQuoted 9 lines
But the point of this argument is:
Get information about NLM (lockd) settings in the current net namespace. This subcommand takes no arguments.
How is that giving information from the running lockd?
What am I missing??
You're missing a kernel that has the required netlink interface. To test this properly, you'll need to patch your kernel, until that patch makes it upstream.
Quoted 8 lines
Quoted 4 lines
Okay... I figured it was something like that. But doesn't make sense to wait until the patch is in upstream so the argument can be properly tested? Why add an argument that will always fail?
Why can't it be properly tested? It's just a matter of running a more recent kernel that has the right interfaces. That should be in linux- next soon (if not already).
I'm doing my testing on a 6.13.0-0.rc6 which will soon be a 6.14 kernel... its my understanding the needed kernel patch will be in the 6.15 kernel... Please correct me if that is not true.
Quoted 4 lines
I think the question is whether we want to wait until the kernel interfaces trickle out into downstream distro kernels before we ship any userland support in an upstream project (nfs-utils).
Yes! As soon as the kernel support hits the upstream kernel, we will be good to go. I just don't want to put a feature in that will fail %100 of the time.
Quoted 5 lines
If you want to wait until it hits Fedora Rawhide kernels, then you're looking at about 10-12 weeks from now. If you want to wait until it makes it into a stable Fedora release kernel then we're looking at about 6 months from now.
nfsdctl is in all current Fedora stable releases, which is the reason I'm pushing back. I do not want to put something in that will make it fail. That just does not make sense to me.
Quoted 1 line
Quoted 9 lines
Quoted 3 lines
I'll note that that it took 6 months to get the original nfsdctl patches merged because of the lag on kernel patches making it into distros, and I think that was way too long.
It took that long because there were issues with the command. In which I was glad to help debug some of the issues...
New technology takes time to develop... I just think this is one of those cases.
Ok, your call. To be clear though, that patch is part of my solution for this bug.
https://issues.redhat.com/browse/RHEL-71698
If you're going to delay it for several months, then can I trouble you to come up with a fix for it that you find acceptable?
Thanks,
Signature · 1 line
Jeff Layton <[email protected]>
Re: [nfs-utils PATCH] nfsdctl: debug logging fixups
Quoted 122 lines · Jeff Layton
Quoted 111 lines · Steve Dickson
Quoted 78 lines · Jeff Layton
Quoted 77 lines · Steve Dickson
Quoted 75 lines · Jeff Layton
Quoted 61 lines · Steve Dickson
Quoted 9 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.Got it...
Quoted 36 lines
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {Ok... at this point we a prettier error message
$ nfsdctl nlm nfsdctl: failed to resolve lockd generic netlink familyQuoted 9 lines
But the point of this argument is:
Get information about NLM (lockd) settings in the current net namespace. This subcommand takes no arguments.
How is that giving information from the running lockd?
What am I missing??
You're missing a kernel that has the required netlink interface. To test this properly, you'll need to patch your kernel, until that patch makes it upstream.
Quoted 8 lines
Quoted 4 lines
Okay... I figured it was something like that. But doesn't make sense to wait until the patch is in upstream so the argument can be properly tested? Why add an argument that will always fail?
Why can't it be properly tested? It's just a matter of running a more recent kernel that has the right interfaces. That should be in linux- next soon (if not already).
I'm doing my testing on a 6.13.0-0.rc6 which will soon be a 6.14 kernel... its my understanding the needed kernel patch will be in the 6.15 kernel... Please correct me if that is not true.
Quoted 4 lines
I think the question is whether we want to wait until the kernel interfaces trickle out into downstream distro kernels before we ship any userland support in an upstream project (nfs-utils).
Yes! As soon as the kernel support hits the upstream kernel, we will be good to go. I just don't want to put a feature in that will fail %100 of the time.
Quoted 5 lines
If you want to wait until it hits Fedora Rawhide kernels, then you're looking at about 10-12 weeks from now. If you want to wait until it makes it into a stable Fedora release kernel then we're looking at about 6 months from now.
nfsdctl is in all current Fedora stable releases, which is the reason I'm pushing back. I do not want to put something in that will make it fail. That just does not make sense to me.
Quoted 1 line
Quoted 9 lines
Quoted 3 lines
I'll note that that it took 6 months to get the original nfsdctl patches merged because of the lag on kernel patches making it into distros, and I think that was way too long.
It took that long because there were issues with the command. In which I was glad to help debug some of the issues...
New technology takes time to develop... I just think this is one of those cases.
Quoted 7 lines
Ok, your call. To be clear though, that patch is part of my solution for this bug.
https://issues.redhat.com/browse/RHEL-71698If you're going to delay it for several months, then can I trouble you to come up with a fix for it that you find acceptable?
How is this a fix when the subcommand will not work without the kernel patch?
I'm sure the subcommand works with the kernel patch but without it... what's the point?
steved.
Jeff Layton
maintainerRe: [nfs-utils PATCH] nfsdctl: debug logging fixups
Quoted 127 lines · Steve Dickson
Quoted 125 lines · Jeff Layton
Quoted 111 lines · Steve Dickson
Quoted 78 lines · Jeff Layton
Quoted 77 lines · Steve Dickson
Quoted 75 lines · Jeff Layton
Quoted 61 lines · Steve Dickson
Quoted 9 lines · Scott Mayhew
Move
read_nfsd_conf()out ofautostart_func()and intomain(). Remove hard-codedNFSD_FAMILY_NAMEin the first error message innetlink_msg_alloc()and make the error messages innetlink_msg_alloc()more descriptive/unique.Signed-off-by Scott Mayhew <[email protected]>---
SteveD - this would go on top of Jeff's "nfsdctl: add support for new lockd configuration interface" patches.Got it...
Quoted 36 lines
utils/nfsdctl/nfsdctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)utils/nfsdctl/nfsdctl.c +2
-4@@ -436,7 +436,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family id =genl_ctrl_resolve (sock, family);if (id <0 ) { -xlog (L_ERROR ,"%s not found" ,NFSD_FAMILY_NAME ); +xlog (L_ERROR ,"failed to resolve %s generic netlink family" , family);return NULL ; }@@ -447,7 +447,7 @@ static struct nl_msg *netlink_msg_alloc(struct nl_sock *sock, const char *family }if (!genlmsg_put (msg,0 ,0 , id,0 ,0 ,0 ,0 )) { -xlog (L_ERROR ,"failed to allocate netlink message" ); +xlog (L_ERROR ,"failed to add generic netlink headers to netlink message" );nlmsg_free (msg);return NULL ; }@@ -1509,8 +1509,6 @@ static int autostart_func(struct nl_sock *sock, int argc, char ** argv) } } -read_nfsd_conf (); - grace =conf_get_num ("nfsd" ,"grace-time" ,0 ); ret =lockd_configure (sock, grace);if (ret) {utils/nfsdctl/nfsdctl.c int main(int argc, char **argv) · line 1733
Quoted 8 lines
utils/nfsdctl/nfsdctl.c +2
-0@@ -1728,6 +1726,8 @@ int main(int argc, char **argv)xlog_syslog (0 );xlog_stderr (1 ); +read_nfsd_conf (); +/* Parse the preliminary options */ while ((opt =getopt_long (argc, argv,"+hdsV" , pre_options,NULL )) !=-1 ) {switch (opt) {Ok... at this point we a prettier error message
$ nfsdctl nlm nfsdctl: failed to resolve lockd generic netlink familyQuoted 9 lines
But the point of this argument is:
Get information about NLM (lockd) settings in the current net namespace. This subcommand takes no arguments.
How is that giving information from the running lockd?
What am I missing??
You're missing a kernel that has the required netlink interface. To test this properly, you'll need to patch your kernel, until that patch makes it upstream.
Quoted 8 lines
Quoted 4 lines
Okay... I figured it was something like that. But doesn't make sense to wait until the patch is in upstream so the argument can be properly tested? Why add an argument that will always fail?
Why can't it be properly tested? It's just a matter of running a more recent kernel that has the right interfaces. That should be in linux- next soon (if not already).
I'm doing my testing on a 6.13.0-0.rc6 which will soon be a 6.14 kernel... its my understanding the needed kernel patch will be in the 6.15 kernel... Please correct me if that is not true.
Quoted 4 lines
I think the question is whether we want to wait until the kernel interfaces trickle out into downstream distro kernels before we ship any userland support in an upstream project (nfs-utils).
Yes! As soon as the kernel support hits the upstream kernel, we will be good to go. I just don't want to put a feature in that will fail %100 of the time.
Quoted 5 lines
If you want to wait until it hits Fedora Rawhide kernels, then you're looking at about 10-12 weeks from now. If you want to wait until it makes it into a stable Fedora release kernel then we're looking at about 6 months from now.
nfsdctl is in all current Fedora stable releases, which is the reason I'm pushing back. I do not want to put something in that will make it fail. That just does not make sense to me.
Quoted 1 line
Quoted 9 lines
Quoted 3 lines
I'll note that that it took 6 months to get the original nfsdctl patches merged because of the lag on kernel patches making it into distros, and I think that was way too long.
It took that long because there were issues with the command. In which I was glad to help debug some of the issues...
New technology takes time to develop... I just think this is one of those cases.
Ok, your call. To be clear though, that patch is part of my solution for this bug.
Quoted 7 lines
Quoted 4 lines
https://issues.redhat.com/browse/RHEL-71698If you're going to delay it for several months, then can I trouble you to come up with a fix for it that you find acceptable?
How is this a fix when the subcommand will not work without the kernel patch?
The nfs-server.service file defines this:
ExecStart=/bin/sh -c '/usr/sbin/nfsdctl autostart || /usr/sbin/rpc.nfsd'
When the lockd netlink interface is needed, but isn't available, then startup will fall back to just calling rpc.nfsd. Currently, the grace_period setting is just ignored, so that fallback just doesn't happen. Very few people will need this; only those that set lockd's ports, or that set the grace_period.
Quoted 2 lines
I'm sure the subcommand works with the kernel patch but without it... what's the point?
Signature · 1 line
Jeff Layton <[email protected]>
Quoted 7 lines · Jeff Layton
v2 is just a small update to fix the problems that Scott spotted.
This patch series adds support for the new lockd configuration interface that should fix this RH bug:
https://issues.redhat.com/browse/RHEL-71698
Quoted 8 lines
There are some other improvements here too, notably a switch to xlog. Only lightly tested, but seems to do the right thing.
Port handling with lockd still needs more work. Currently that is usually configured by rpc.statd. I think we need to convert it to use netlink to configure the ports as well, when it's able.
Signed-off-by Jeff Layton <[email protected]>
I have reluctantly committed these because the
nlm arg still fails (genl_ctrl_resolve() is not finding the family) but all the other args seem to work and other patches are dependent on these that do fixes bugs.
The Bakeathon is coming up so hopefully we can test these patch there.
Committed (tag: nfs-utils-2-8-3-rc7)
steved.
Quoted 24 lines
---
Changes in v2:
- properly regenerate manpages
- fix up bogus merge conflict
- addD_GENERALxlog messages when nfsdctl starts and exits
- Link to v1: https://lore.kernel.org/r/[email protected]---
Jeff Layton (3):
nfsdctl: convert toxlog()
nfsdctl: fix the --version option
nfsdctl: add necessary bits to configure lockdconfigure.ac | 4 + utils/nfsdctl/lockd_netlink.h | 29 ++++ utils/nfsdctl/nfsdctl.8 | 15 +- utils/nfsdctl/nfsdctl.adoc | 8 + utils/nfsdctl/nfsdctl.c | 331 ++++++++++++++++++++++++++++++++++-------- 5 files changed, 321 insertions(+), 66 deletions(-)---
base-commit:65f4cc3a6ce1472ee4092c4bbf4b19beb0a8217b
change-id: 20250109-lockd-nl-6272fa9e8a5dBest regards,