Make nfs and nfsd stats visible in network ns
Subsystems (3)
v2 cover Make nfs and nfsd stats visible in network ns
v1: https://lore.kernel.org/linux-nfs/[email protected]/
v1->v2:
- rework the sunprc service creation to take a pointer to the sv_stats.
- dropped ->pg_stats from the svc_program.
- converted all of the nfsd global stats to per-network namespace.
- added the ability to point at a specific rpc_stat for rpc program creation.
- converted the rpc stats for nfs to per-network namespace.
-- Original email --
Hello,
We're currently deploying NFS internally and have run into some oddities with our usage of containers. All of the services that mount and export NFS volumes run inside of containers, specifically all the namespaces including network namespaces. Our monitoring is done on a per-container basis, so we need access to the nfs and nfsd stats that are under /proc/net/sunrpc. However these are only tied to the init_net, which makes them invisible to containers in a different network namespace.
Fix this so that these files are tied to the network namespace. This allows us to avoid the hack of bind mounting the hosts /proc into the container in order to do proper monitoring. Thanks,
Josef
Josef Bacik (13):
sunrpc: don't change ->sv_stats if it doesn't exist
nfs: stop setting ->pg_stats for unused stats
sunrpc: pass in the sv_stats struct through svc_create*
sunrpc: remove ->pg_stats from svc_program
sunrpc: add a struct rpc_stats arg to rpc_create_args
sunrpc: use the struct net as the svc proc private
nfsd: rename NFSD_NET_* to NFSD_STATS_*
nfsd: expose /proc/net/sunrpc/nfsd in net namespaces
nfsd: make all of the nfsd stats per-network namespace
nfsd: move th_cnt into nfsd_net
nfsd: make svc_stat per-network namespace instead of global
nfs: expose /proc/net/sunrpc/nfs in net namespaces
nfs: make the rpc_stat per net namespace
fs/lockd/svc.c | 5 +-- fs/nfs/callback.c | 5 +-- fs/nfs/client.c | 5 ++- fs/nfs/inode.c | 8 ++--- fs/nfs/internal.h | 2 -- fs/nfs/netns.h | 2 ++ fs/nfsd/cache.h | 2 -- fs/nfsd/netns.h | 28 ++++++++++++--- fs/nfsd/nfs4proc.c | 6 ++-- fs/nfsd/nfs4state.c | 3 +- fs/nfsd/nfscache.c | 40 +++++---------------- fs/nfsd/nfsctl.c | 16 ++++----- fs/nfsd/nfsfh.c | 3 +- fs/nfsd/nfssvc.c | 13 +++---- fs/nfsd/stats.c | 53 ++++++++++++---------------- fs/nfsd/stats.h | 70 +++++++++++++------------------------ fs/nfsd/vfs.c | 5 +-- include/linux/sunrpc/clnt.h | 1 + include/linux/sunrpc/svc.h | 9 +++-- net/sunrpc/clnt.c | 2 +- net/sunrpc/stats.c | 2 +- net/sunrpc/svc.c | 44 ++++++++++++++--------- 22 files changed, 147 insertions(+), 177 deletions(-)
Signature · 1 line
2.43.0
PATCH v2 1/13 sunrpc: don't change ->sv_stats if it doesn't exist
We check for the existence of ->sv_stats elsewhere except in the core processing code. It appears that only nfsd actual exports these values anywhere, everybody else just has a write only copy of sv_stats in their svc_program. Add a check for ->sv_stats before every adjustment to allow us to eliminate the stats struct from all the users who don't report the stats.
--- net/sunrpc/svc.c | 27 ++++++++++++++++++--------- 1 file changed, 18 insertions(+), 9 deletions(-)
net/sunrpc/svc.c +18 -9
@@ -1375,7 +1375,8 @@ svc_process_common(struct svc_rqst *rqstp) goto err_bad_proc;/* Syntactic check complete */ - serv->sv_stats->rpccnt++; +if (serv->sv_stats) + serv->sv_stats->rpccnt++;trace_svc_process (rqstp, progp->pg_name); aoffset =xdr_stream_pos (xdr);
@@ -1427,7 +1428,8 @@ svc_process_common(struct svc_rqst *rqstp) goto close_xprt; err_bad_rpc: - serv->sv_stats->rpcbadfmt++; +if (serv->sv_stats) + serv->sv_stats->rpcbadfmt++;xdr_stream_encode_u32 (xdr,RPC_MSG_DENIED );xdr_stream_encode_u32 (xdr,RPC_MISMATCH );/* Only RPCv2 supported */
@@ -1438,7 +1440,8 @@ svc_process_common(struct svc_rqst *rqstp) err_bad_auth:dprintk ("svc: authentication failed (%d)\n" ,be32_to_cpu (rqstp->rq_auth_stat)); - serv->sv_stats->rpcbadauth++; +if (serv->sv_stats) + serv->sv_stats->rpcbadauth++;/* Restore write pointer to location of reply status: */ xdr_truncate_encode (xdr,XDR_UNIT *2 );xdr_stream_encode_u32 (xdr,RPC_MSG_DENIED );
@@ -1448,7 +1451,8 @@ svc_process_common(struct svc_rqst *rqstp) err_bad_prog:dprintk ("svc: unknown program %d\n" , rqstp->rq_prog); - serv->sv_stats->rpcbadfmt++; +if (serv->sv_stats) + serv->sv_stats->rpcbadfmt++; *rqstp->rq_accept_statp = rpc_prog_unavail; goto sendit;
@@ -1456,7 +1460,8 @@ svc_process_common(struct svc_rqst *rqstp)svc_printk (rqstp,"unknown version (%d for prog %d, %s)\n" , rqstp->rq_vers, rqstp->rq_prog, progp->pg_name); - serv->sv_stats->rpcbadfmt++; +if (serv->sv_stats) + serv->sv_stats->rpcbadfmt++; *rqstp->rq_accept_statp = rpc_prog_mismatch;/*
@@ -1470,19 +1475,22 @@ svc_process_common(struct svc_rqst *rqstp) err_bad_proc:svc_printk (rqstp,"unknown procedure (%d)\n" , rqstp->rq_proc); - serv->sv_stats->rpcbadfmt++; +if (serv->sv_stats) + serv->sv_stats->rpcbadfmt++; *rqstp->rq_accept_statp = rpc_proc_unavail; goto sendit; err_garbage_args:svc_printk (rqstp,"failed to decode RPC header\n" ); - serv->sv_stats->rpcbadfmt++; +if (serv->sv_stats) + serv->sv_stats->rpcbadfmt++; *rqstp->rq_accept_statp = rpc_garbage_args; goto sendit; err_system_err: - serv->sv_stats->rpcbadfmt++; +if (serv->sv_stats) + serv->sv_stats->rpcbadfmt++; *rqstp->rq_accept_statp = rpc_system_err; goto sendit; }
@@ -1534,7 +1542,8 @@ void svc_process(struct svc_rqst *rqstp) out_baddir:svc_printk (rqstp,"bad direction 0x%08x, dropping request\n" ,be32_to_cpu (*p)); - rqstp->rq_server->sv_stats->rpcbadfmt++; +if (rqstp->rq_server->sv_stats) + rqstp->rq_server->sv_stats->rpcbadfmt++; out_drop:svc_drop (rqstp); }
Signature · 1 line
2.43.0
PATCH v2 2/13 nfs: stop setting ->pg_stats for unused stats
In reply to Josef BacikA lot of places are setting a blank svc_stats in ->pg_stats and never utilizing these stats. Remove all of these extra structs as we're not reporting these stats anywhere.
--- fs/lockd/svc.c | 3 --- fs/nfs/callback.c | 3 --- fs/nfsd/nfssvc.c | 5 ----- 3 files changed, 11 deletions(-)
fs/lockd/svc.c +0 -3
@@ -710,8 +710,6 @@ static const struct svc_version *nlmsvc_version[] = {#endif }; -static struct svc_stat nlmsvc_stats; -#define NLM_NRVERS ARRAY_SIZE(nlmsvc_version)static struct svc_program nlmsvc_program = { .pg_prog =NLM_PROGRAM ,/* program number */
@@ -719,7 +717,6 @@ static struct svc_program nlmsvc_program = { .pg_vers = nlmsvc_version, /* version table */ .pg_name = "lockd", /* service name */ .pg_class = "nfsd", /* share authentication with nfsd */ - .pg_stats = &nlmsvc_stats, /* stats table */ .pg_authenticate = &lockd_authenticate, /* export authentication */ .pg_init_request = svc_generic_init_request, .pg_rpcbind_set = svc_generic_rpcbind_set,
fs/nfs/callback.c +0 -3
@@ -356,15 +356,12 @@ static const struct svc_version *nfs4_callback_version[] = { [4] = &nfs4_callback_version4, }; -static struct svc_stat nfs4_callback_stats; -static struct svc_program nfs4_callback_program = { .pg_prog =NFS4_CALLBACK ,/* RPC service number */ .pg_nvers =ARRAY_SIZE (nfs4_callback_version),/* Number of entries */ .pg_vers = nfs4_callback_version,/* version table */ .pg_name ="NFSv4 callback" ,/* service name */ .pg_class ="nfs" ,/* authentication class */ - .pg_stats = &nfs4_callback_stats, .pg_authenticate = nfs_callback_authenticate, .pg_init_request = svc_generic_init_request, .pg_rpcbind_set = svc_generic_rpcbind_set,
fs/nfsd/nfssvc.c +0 -5
@@ -80,7 +80,6 @@ unsigned long nfsd_drc_max_mem;unsigned long nfsd_drc_mem_used;#if defined(CONFIG_NFSD_V2_ACL ) || defined(CONFIG_NFSD_V3_ACL ) -static struct svc_stat nfsd_acl_svcstats;static const struct svc_version *nfsd_acl_version[] = { # if defined(CONFIG_NFSD_V2_ACL) [2] = &nfsd_acl_version2,
@@ -99,15 +98,11 @@ static struct svc_program nfsd_acl_program = { .pg_vers = nfsd_acl_version, .pg_name = "nfsacl", .pg_class = "nfsd", - .pg_stats = &nfsd_acl_svcstats, .pg_authenticate = &svc_set_client, .pg_init_request = nfsd_acl_init_request, .pg_rpcbind_set = nfsd_acl_rpcbind_set, }; -static struct svc_stat nfsd_acl_svcstats = { - .program = &nfsd_acl_program, -};#endif /* defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL) */ static const struct svc_version *nfsd_version[] = {
Signature · 1 line
2.43.0
PATCH v2 3/13 sunrpc: pass in the sv_stats struct through svc_create*
In reply to Josef BacikSince only one service actually reports the rpc stats there's not much of a reason to have a pointer to it in the svc_program struct. Adjust the svc_create* functions to take the sv_stats as an argument and pass the struct through there as desired instead of getting it from the svc_program->pg_stats.
--- fs/lockd/svc.c | 2 +- fs/nfs/callback.c | 2 +- fs/nfsd/nfssvc.c | 3 ++- include/linux/sunrpc/svc.h | 8 ++++---- net/sunrpc/svc.c | 17 ++++++++++------- 5 files changed, 18 insertions(+), 14 deletions(-)
fs/lockd/svc.c +1 -1
@@ -337,7 +337,7 @@ static int lockd_get(void) nlm_timeout =LOCKD_DFLT_TIMEO ; nlmsvc_timeout = nlm_timeout *HZ ; - serv =svc_create (&nlmsvc_program,LOCKD_BUFSIZE , lockd); + serv =svc_create (&nlmsvc_program,NULL ,LOCKD_BUFSIZE , lockd);if (!serv) {printk (KERN_WARNING "lockd_up: create service failed\n" );return -ENOMEM ;
fs/nfs/callback.c +1 -1
@@ -202,7 +202,7 @@ static struct svc_serv *nfs_callback_create_svc(int minorversion)if (minorversion)return ERR_PTR (-ENOTSUPP );#endif - serv =svc_create (&nfs4_callback_program,NFS4_CALLBACK_BUFSIZE , + serv =svc_create (&nfs4_callback_program,NULL ,NFS4_CALLBACK_BUFSIZE , threadfn);if (!serv) {printk (KERN_ERR "nfs_callback_create_svc: create service failed\n" );
fs/nfsd/nfssvc.c +2 -1
@@ -661,7 +661,8 @@ int nfsd_create_serv(struct net *net)if (nfsd_max_blksize ==0 ) nfsd_max_blksize =nfsd_get_default_max_blksize ();nfsd_reset_versions (nn); - serv =svc_create_pooled (&nfsd_program, nfsd_max_blksize, nfsd); + serv =svc_create_pooled (&nfsd_program, &nfsd_svcstats, + nfsd_max_blksize, nfsd);if (serv ==NULL )return -ENOMEM ;
include/linux/sunrpc/svc.h +4 -4
@@ -402,8 +402,8 @@ struct svc_procedure {int svc_rpcb_setup (struct svc_serv *serv,struct net *net);void svc_rpcb_cleanup (struct svc_serv *serv,struct net *net);int svc_bind (struct svc_serv *serv,struct net *net); -struct svc_serv *svc_create (struct svc_program *,unsigned int , -int (*threadfn)(void *data)); +struct svc_serv *svc_create (struct svc_program *,struct svc_stat *, +unsigned int ,int (*threadfn)(void *data));struct svc_rqst *svc_rqst_alloc (struct svc_serv *serv,struct svc_pool *pool,int node); bool svc_rqst_replace_page(struct svc_rqst *rqstp,
@@ -411,8 +411,8 @@ bool svc_rqst_replace_page(struct svc_rqst *rqstp,void svc_rqst_release_pages (struct svc_rqst *rqstp);void svc_rqst_free (struct svc_rqst *);void svc_exit_thread (struct svc_rqst *); -struct svc_serv *svc_create_pooled (struct svc_program *,unsigned int , -int (*threadfn)(void *data)); +struct svc_serv *svc_create_pooled (struct svc_program *,struct svc_stat *, +unsigned int ,int (*threadfn)(void *data));int svc_set_num_threads (struct svc_serv *,struct svc_pool *,int );int svc_pool_stats_open (struct svc_info *si,struct file *file);void svc_process (struct svc_rqst *rqstp);
net/sunrpc/svc.c +10 -7
@@ -451,8 +451,8 @@ __svc_init_bc(struct svc_serv *serv)* Create an RPC service */ static struct svc_serv * -__svc_create (struct svc_program *prog,unsigned int bufsize,int npools, -int (*threadfn)(void *data)) +__svc_create (struct svc_program *prog,struct svc_stat *stats, +unsigned int bufsize,int npools,int (*threadfn)(void *data)) {struct svc_serv *serv;unsigned int vers;
@@ -463,7 +463,7 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,return NULL ; serv->sv_name = prog->pg_name; serv->sv_program = prog; - serv->sv_stats = prog->pg_stats; + serv->sv_stats = stats;if (bufsize >RPCSVC_MAXPAYLOAD ) bufsize =RPCSVC_MAXPAYLOAD ; serv->sv_max_payload = bufsize? bufsize :4096 ;
@@ -521,34 +521,37 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,/** * svc_create - Create an RPC service * @prog: the RPC program the new service will handle +* @stats: the stats struct if desired * @bufsize: maximum message size for @prog * @threadfn: a function to service RPC requests for @prog * * Returns an instantiated struct svc_serv object or NULL. */ -struct svc_serv *svc_create (struct svc_program *prog,unsigned int bufsize, -int (*threadfn)(void *data)) +struct svc_serv *svc_create (struct svc_program *prog,struct svc_stat *stats, +unsigned int bufsize,int (*threadfn)(void *data)) { -return __svc_create (prog, bufsize,1 , threadfn); +return __svc_create (prog, stats, bufsize,1 , threadfn); }EXPORT_SYMBOL_GPL (svc_create);/** * svc_create_pooled - Create an RPC service with pooled threads * @prog: the RPC program the new service will handle +* @stats: the stats struct if desired * @bufsize: maximum message size for @prog * @threadfn: a function to service RPC requests for @prog * * Returns an instantiated struct svc_serv object or NULL. */ struct svc_serv *svc_create_pooled (struct svc_program *prog, +struct svc_stat *stats,unsigned int bufsize,int (*threadfn)(void *data)) {struct svc_serv *serv;unsigned int npools =svc_pool_map_get (); - serv =__svc_create (prog, bufsize, npools, threadfn); + serv =__svc_create (prog, stats, bufsize, npools, threadfn);if (!serv) goto out_err;return serv;
Signature · 1 line
2.43.0
PATCH v2 4/13 sunrpc: remove ->pg_stats from svc_program
In reply to Josef BacikNow that this isn't used anywhere, remove it.
--- fs/nfsd/nfssvc.c | 1 - include/linux/sunrpc/svc.h | 1 - 2 files changed, 2 deletions(-)
fs/nfsd/nfssvc.c +0 -1
@@ -127,7 +127,6 @@ struct svc_program nfsd_program = { .pg_vers = nfsd_version, /* version table */ .pg_name = "nfsd", /* program name */ .pg_class = "nfsd", /* authentication class */ - .pg_stats = &nfsd_svcstats, /* version table */ .pg_authenticate = &svc_set_client, /* export authentication */ .pg_init_request = nfsd_init_request, .pg_rpcbind_set = nfsd_rpcbind_set,
include/linux/sunrpc/svc.h +0 -1
@@ -339,7 +339,6 @@ struct svc_program {const struct svc_version **pg_vers;/* version array */ char * pg_name;/* service name */ char * pg_class;/* class name: services sharing authentication */ -struct svc_stat * pg_stats;/* rpc statistics */ enum svc_auth_status (*pg_authenticate)(struct svc_rqst *rqstp); __be32 (*pg_init_request)(struct svc_rqst *, const struct svc_program *,
Signature · 1 line
2.43.0
PATCH v2 5/13 sunrpc: add a struct rpc_stats arg to rpc_create_args
In reply to Josef BacikWe want to be able to have our rpc stats handled in a per network namespace manner, so add an option to rpc_create_args to specify a different rpc_stats struct instead of using the one on the rpc_program.
--- fs/nfs/client.c | 2 +- include/linux/sunrpc/clnt.h | 1 + net/sunrpc/clnt.c | 2 +- 3 files changed, 3 insertions(+), 2 deletions(-)
fs/nfs/client.c +1 -1
@@ -73,7 +73,7 @@ const struct rpc_program nfs_program = { .number = NFS_PROGRAM, .nrvers = ARRAY_SIZE(nfs_version), .version = nfs_version, - .stats = &nfs_rpcstat, + .stats = &nfs_rpcstat, .pipe_dir_name = NFS_PIPE_DIRNAME, };
include/linux/sunrpc/clnt.h +1 -0
@@ -139,6 +139,7 @@ struct rpc_create_args {const char *servername;const char *nodename;const struct rpc_program *program; +struct rpc_stat *stats;u32 prognumber;/* overrides program->number */ u32 version;rpc_authflavor_t authflavor;
net/sunrpc/clnt.c +1 -1
@@ -405,7 +405,7 @@ static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args, clnt->cl_maxproc = version->nrprocs; clnt->cl_prog = args->prognumber ? : program->number; clnt->cl_vers = version->number; - clnt->cl_stats = program->stats; + clnt->cl_stats = args->stats ? : program->stats; clnt->cl_metrics =rpc_alloc_iostats (clnt);rpc_init_pipe_dir_head (&clnt->cl_pipedir_objects); err = -ENOMEM ;
Signature · 1 line
2.43.0
PATCH v2 6/13 sunrpc: use the struct net as the svc proc private
In reply to Josef Baciknfsd is the only thing using this helper, and it doesn't use the private currently. When we switch to per-network namespace stats we will need the struct net * in order to get to the nfsd_net. Use the net as the proc private so we can utilize this when we make the switch over.
--- net/sunrpc/stats.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
net/sunrpc/stats.c +1 -1
@@ -314,7 +314,7 @@ EXPORT_SYMBOL_GPL(rpc_proc_unregister);struct proc_dir_entry *svc_proc_register (struct net *net,struct svc_stat *statp,const struct proc_ops *proc_ops) { -return do_register (net, statp->program->pg_name, statp, proc_ops); +return do_register (net, statp->program->pg_name, net, proc_ops); }EXPORT_SYMBOL_GPL (svc_proc_register);
Signature · 1 line
2.43.0
PATCH v2 7/13 nfsd: rename NFSD_NET_* to NFSD_STATS_*
In reply to Josef BacikWe're going to merge the stats all into per network namespace in subsequent patches, rename these nn counters to be consistent with the rest of the stats.
--- fs/nfsd/netns.h | 4 ++-- fs/nfsd/nfscache.c | 4 ++-- fs/nfsd/stats.h | 6 +++--- 3 files changed, 7 insertions(+), 7 deletions(-)
fs/nfsd/netns.h +2 -2
@@ -26,9 +26,9 @@ struct nfsd4_client_tracking_ops;enum {/* cache misses due only to checksum comparison failures */ -NFSD_NET_PAYLOAD_MISSES , +NFSD_STATS_PAYLOAD_MISSES ,/* amount of memory (in bytes) currently consumed by the DRC */ -NFSD_NET_DRC_MEM_USAGE , +NFSD_STATS_DRC_MEM_USAGE ,NFSD_NET_COUNTERS_NUM };
fs/nfsd/nfscache.c +2 -2
@@ -687,7 +687,7 @@ int nfsd_reply_cache_stats_show(struct seq_file *m, void *v) atomic_read(&nn->num_drc_entries));seq_printf (m,"hash buckets: %u\n" ,1 << nn->maskbits);seq_printf (m,"mem usage: %lld\n" , -percpu_counter_sum_positive (&nn->counter[NFSD_NET_DRC_MEM_USAGE ])); +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_DRC_MEM_USAGE ]));seq_printf (m,"cache hits: %lld\n" ,percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_RC_HITS ])); seq_printf(m, "cache misses: %lld\n",
@@ -695,7 +695,7 @@ int nfsd_reply_cache_stats_show(struct seq_file *m, void *v)seq_printf (m,"not cached: %lld\n" ,percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_RC_NOCACHE ]));seq_printf (m,"payload misses: %lld\n" , -percpu_counter_sum_positive (&nn->counter[NFSD_NET_PAYLOAD_MISSES ])); +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_PAYLOAD_MISSES ]));seq_printf (m,"longest chain len: %u\n" , nn->longest_chain);seq_printf (m,"cachesize at longest: %u\n" , nn->longest_chain_cachesize);return 0 ;
fs/nfsd/stats.h +3 -3
@@ -81,17 +81,17 @@ static inline void nfsd_stats_io_write_add(struct svc_export *exp, s64 amount)static inline void nfsd_stats_payload_misses_inc (struct nfsd_net *nn) { -percpu_counter_inc (&nn->counter[NFSD_NET_PAYLOAD_MISSES ]); +percpu_counter_inc (&nn->counter[NFSD_STATS_PAYLOAD_MISSES ]); }static inline void nfsd_stats_drc_mem_usage_add (struct nfsd_net *nn,s64 amount) { -percpu_counter_add (&nn->counter[NFSD_NET_DRC_MEM_USAGE ], amount); +percpu_counter_add (&nn->counter[NFSD_STATS_DRC_MEM_USAGE ], amount); }static inline void nfsd_stats_drc_mem_usage_sub (struct nfsd_net *nn,s64 amount) { -percpu_counter_sub (&nn->counter[NFSD_NET_DRC_MEM_USAGE ], amount); +percpu_counter_sub (&nn->counter[NFSD_STATS_DRC_MEM_USAGE ], amount); }#ifdef CONFIG_NFSD_V4
Signature · 1 line
2.43.0
PATCH v2 8/13 nfsd: expose /proc/net/sunrpc/nfsd in net namespaces
In reply to Josef BacikWe are running nfsd servers inside of containers with their own network namespace, and we want to monitor these services using the stats found in /proc. However these are not exposed in the proc inside of the container, so we have to bind mount the host /proc into our containers to get at this information.
Separate out the stat counters init and the proc registration, and move the proc registration into the pernet operations entry and exit points so that these stats can be exposed inside of network namespaces.
This is an intermediate step, this just exposes the global counters in the network namespace. Subsequent patches will move these counters into the per-network namespace container.
--- fs/nfsd/nfsctl.c | 8 +++++--- fs/nfsd/stats.c | 21 ++++++--------------- fs/nfsd/stats.h | 6 ++++-- 3 files changed, 15 insertions(+), 20 deletions(-)
fs/nfsd/nfsctl.c +5 -3
@@ -1679,6 +1679,7 @@ static __net_init int nfsd_net_init(struct net *net)nfsd4_init_leases_net (nn);get_random_bytes (&nn->siphash_key,sizeof (nn->siphash_key));seqlock_init (&nn->writeverf_lock); +nfsd_proc_stat_init (net);return 0 ;
@@ -1699,6 +1700,7 @@ static __net_exit void nfsd_net_exit(struct net *net) {struct nfsd_net *nn =net_generic (net, nfsd_net_id); +nfsd_proc_stat_shutdown (net);nfsd_net_reply_cache_destroy (nn);nfsd_idmap_shutdown (net);nfsd_export_shutdown (net);
@@ -1722,7 +1724,7 @@ static int __init init_nfsd(void) retval =nfsd4_init_pnfs ();if (retval) goto out_free_slabs; - retval =nfsd_stat_init ();/* Statistics */ + retval =nfsd_stat_counters_init ();/* Statistics */ if (retval) goto out_free_pnfs; retval =nfsd_drc_slab_create ();
@@ -1762,7 +1764,7 @@ static int __init init_nfsd(void)nfsd_lockd_shutdown ();nfsd_drc_slab_free (); out_free_stat: -nfsd_stat_shutdown (); +nfsd_stat_counters_destroy (); out_free_pnfs:nfsd4_exit_pnfs (); out_free_slabs:
@@ -1780,7 +1782,7 @@ static void __exit exit_nfsd(void)nfsd_drc_slab_free ();remove_proc_entry ("fs/nfs/exports" ,NULL );remove_proc_entry ("fs/nfs" ,NULL ); -nfsd_stat_shutdown (); +nfsd_stat_counters_destroy ();nfsd_lockd_shutdown ();nfsd4_free_slabs ();nfsd4_exit_pnfs ();
fs/nfsd/stats.c +6 -15
@@ -108,31 +108,22 @@ void nfsd_percpu_counters_destroy(struct percpu_counter counters[], int num)percpu_counter_destroy (&counters[i]); } -static int nfsd_stat_counters_init (void ) +int nfsd_stat_counters_init (void ) {return nfsd_percpu_counters_init (nfsdstats.counter,NFSD_STATS_COUNTERS_NUM ); } -static void nfsd_stat_counters_destroy (void ) +void nfsd_stat_counters_destroy (void ) {nfsd_percpu_counters_destroy (nfsdstats.counter,NFSD_STATS_COUNTERS_NUM ); } -int nfsd_stat_init (void ) +void nfsd_proc_stat_init (struct net *net) { -int err; - - err =nfsd_stat_counters_init (); -if (err) -return err; - -svc_proc_register (&init_net, &nfsd_svcstats, &nfsd_proc_ops); - -return 0 ; +svc_proc_register (net, &nfsd_svcstats, &nfsd_proc_ops); } -void nfsd_stat_shutdown (void ) +void nfsd_proc_stat_shutdown (struct net *net) { -nfsd_stat_counters_destroy (); -svc_proc_unregister (&init_net,"nfsd" ); +svc_proc_unregister (net,"nfsd" ); }
fs/nfsd/stats.h +4 -2
@@ -40,8 +40,10 @@ extern struct svc_stat nfsd_svcstats;int nfsd_percpu_counters_init (struct percpu_counter *counters,int num);void nfsd_percpu_counters_reset (struct percpu_counter *counters,int num);void nfsd_percpu_counters_destroy (struct percpu_counter *counters,int num); -int nfsd_stat_init (void ); -void nfsd_stat_shutdown (void ); +int nfsd_stat_counters_init (void ); +void nfsd_stat_counters_destroy (void ); +void nfsd_proc_stat_init (struct net *net); +void nfsd_proc_stat_shutdown (struct net *net);static inline void nfsd_stats_rc_hits_inc (void ) {
Signature · 1 line
2.43.0
PATCH v2 9/13 nfsd: make all of the nfsd stats per-network namespace
In reply to Josef BacikWe have a global set of counters that we modify for all of the nfsd operations, but now that we're exposing these stats across all network namespaces we need to make the stats also be per-network namespace. We already have some caching stats that are per-network namespace, so move these definitions into the same counter and then adjust all the helpers and users of these stats to provide the appropriate nfsd_net struct so that the stats are maintained for the per-network namespace objects.
--- fs/nfsd/cache.h | 2 -- fs/nfsd/netns.h | 17 ++++++++++++-- fs/nfsd/nfs4proc.c | 6 ++--- fs/nfsd/nfs4state.c | 3 ++- fs/nfsd/nfscache.c | 36 ++++++------------------------ fs/nfsd/nfsctl.c | 12 +++------- fs/nfsd/nfsfh.c | 3 ++- fs/nfsd/stats.c | 26 ++++++++++++---------- fs/nfsd/stats.h | 54 ++++++++++++++++----------------------------- fs/nfsd/vfs.c | 5 +++-- 10 files changed, 68 insertions(+), 96 deletions(-)
fs/nfsd/cache.h +0 -2
@@ -80,8 +80,6 @@ enum {int nfsd_drc_slab_create (void );void nfsd_drc_slab_free (void ); -int nfsd_net_reply_cache_init (struct nfsd_net *nn); -void nfsd_net_reply_cache_destroy (struct nfsd_net *nn);int nfsd_reply_cache_init (struct nfsd_net *);void nfsd_reply_cache_shutdown (struct nfsd_net *); int nfsd_cache_lookup(struct svc_rqst *rqstp, unsigned int start,
fs/nfsd/netns.h +15 -2
@@ -11,6 +11,7 @@#include <net/net_namespace.h> #include <net/netns/generic.h> #include <linux/filelock.h> +#include <linux/nfs4.h> #include <linux/percpu_counter.h> #include <linux/siphash.h>
@@ -29,7 +30,19 @@ enum {NFSD_STATS_PAYLOAD_MISSES ,/* amount of memory (in bytes) currently consumed by the DRC */ NFSD_STATS_DRC_MEM_USAGE , - NFSD_NET_COUNTERS_NUM +NFSD_STATS_RC_HITS ,/* repcache hits */ +NFSD_STATS_RC_MISSES ,/* repcache misses */ +NFSD_STATS_RC_NOCACHE ,/* uncached reqs */ +NFSD_STATS_FH_STALE ,/* FH stale error */ +NFSD_STATS_IO_READ ,/* bytes returned to read requests */ +NFSD_STATS_IO_WRITE ,/* bytes passed in write requests */ +#ifdef CONFIG_NFSD_V4 +NFSD_STATS_FIRST_NFS4_OP ,/* count of individual nfsv4 operations */ +NFSD_STATS_LAST_NFS4_OP =NFSD_STATS_FIRST_NFS4_OP +LAST_NFS4_OP , +#define NFSD_STATS_NFS4_OP(op) (NFSD_STATS_FIRST_NFS4_OP + (op)) +NFSD_STATS_WDELEG_GETATTR ,/* count of getattr conflict with wdeleg */ +#endif +NFSD_STATS_COUNTERS_NUM };/*
@@ -164,7 +177,7 @@ struct nfsd_net {atomic_t num_drc_entries;/* Per-netns stats counters */ -struct percpu_counter counter[NFSD_NET_COUNTERS_NUM ]; +struct percpu_counter counter[NFSD_STATS_COUNTERS_NUM ];/* longest hash chain seen */ unsigned int longest_chain;
fs/nfsd/nfs4proc.c +3 -3
@@ -2490,10 +2490,10 @@ nfsd4_proc_null(struct svc_rqst *rqstp)return rpc_success; } -static inline void nfsd4_increment_op_stats (u32 opnum) +static inline void nfsd4_increment_op_stats (struct nfsd_net *nn,u32 opnum) {if (opnum >=FIRST_NFS4_OP && opnum <=LAST_NFS4_OP ) -percpu_counter_inc (&nfsdstats.counter[NFSD_STATS_NFS4_OP (opnum)]); +percpu_counter_inc (&nn->counter[NFSD_STATS_NFS4_OP (opnum)]); }static const struct nfsd4_operation nfsd4_ops[];
@@ -2768,7 +2768,7 @@ nfsd4_proc_compound(struct svc_rqst *rqstp) status, nfsd4_op_name(op->opnum));nfsd4_cstate_clear_replay (cstate); -nfsd4_increment_op_stats (op->opnum); +nfsd4_increment_op_stats (nn, op->opnum); }fh_put (current_fh);
fs/nfsd/nfs4state.c +2 -1
@@ -8457,6 +8457,7 @@ __be32 nfsd4_deleg_getattr_conflict(struct svc_rqst *rqstp, struct inode *inode) { __be32 status; +struct nfsd_net *nn =net_generic (SVC_NET (rqstp), nfsd_net_id);struct file_lock_context *ctx;struct file_lock *fl;struct nfs4_delegation *dp;
@@ -8486,7 +8487,7 @@ nfsd4_deleg_getattr_conflict(struct svc_rqst *rqstp, struct inode *inode) } break_lease:spin_unlock (&ctx->flc_lock); -nfsd_stats_wdeleg_getattr_inc (); +nfsd_stats_wdeleg_getattr_inc (nn); status =nfserrno (nfsd_open_break_lease (inode,NFSD_MAY_READ ));if (status != nfserr_jukebox || !nfsd_wait_for_delegreturn (rqstp, inode))
fs/nfsd/nfscache.c +7 -29
@@ -176,27 +176,6 @@ void nfsd_drc_slab_free(void)kmem_cache_destroy (drc_slab); } -/** -* nfsd_net_reply_cache_init - per net namespace reply cache set-up -* @nn: nfsd_net being initialized -* -* Returns zero on succes; otherwise a negative errno is returned. -*/ -int nfsd_net_reply_cache_init (struct nfsd_net *nn) -{ -return nfsd_percpu_counters_init (nn->counter,NFSD_NET_COUNTERS_NUM ); -} - -/** -* nfsd_net_reply_cache_destroy - per net namespace reply cache tear-down -* @nn: nfsd_net being freed -* -*/ -void nfsd_net_reply_cache_destroy (struct nfsd_net *nn) -{ -nfsd_percpu_counters_destroy (nn->counter,NFSD_NET_COUNTERS_NUM ); -} -int nfsd_reply_cache_init (struct nfsd_net *nn) {unsigned int hashsize;
@@ -501,7 +480,7 @@ nfsd_cache_insert(struct nfsd_drc_bucket *b, struct nfsd_cacherep *key,int nfsd_cache_lookup (struct svc_rqst *rqstp,unsigned int start,unsigned int len,struct nfsd_cacherep **cacherep) { -struct nfsd_net *nn; +struct nfsd_net *nn =net_generic (SVC_NET (rqstp), nfsd_net_id);struct nfsd_cacherep *rp, *found;__wsum csum;struct nfsd_drc_bucket *b;
@@ -510,7 +489,7 @@ int nfsd_cache_lookup(struct svc_rqst *rqstp, unsigned int start,int rtn =RC_DOIT ;if (type ==RC_NOCACHE ) { -nfsd_stats_rc_nocache_inc (); +nfsd_stats_rc_nocache_inc (nn); goto out; }
@@ -520,7 +499,6 @@ int nfsd_cache_lookup(struct svc_rqst *rqstp, unsigned int start,* Since the common case is a cache miss followed by an insert, * preallocate an entry. */ - nn =net_generic (SVC_NET (rqstp), nfsd_net_id); rp =nfsd_cacherep_alloc (rqstp, csum, nn);if (!rp) goto out;
@@ -537,7 +515,7 @@ int nfsd_cache_lookup(struct svc_rqst *rqstp, unsigned int start,nfsd_cacherep_dispose (&dispose); -nfsd_stats_rc_misses_inc (); +nfsd_stats_rc_misses_inc (nn);atomic_inc (&nn->num_drc_entries);nfsd_stats_drc_mem_usage_add (nn,sizeof (*rp)); goto out;
@@ -545,7 +523,7 @@ int nfsd_cache_lookup(struct svc_rqst *rqstp, unsigned int start, found_entry:/* We found a matching entry which is either in progress or done. */ nfsd_reply_cache_free_locked (NULL , rp, nn); -nfsd_stats_rc_hits_inc (); +nfsd_stats_rc_hits_inc (nn); rtn =RC_DROPIT ; rp = found;
@@ -689,11 +667,11 @@ int nfsd_reply_cache_stats_show(struct seq_file *m, void *v)seq_printf (m,"mem usage: %lld\n" ,percpu_counter_sum_positive (&nn->counter[NFSD_STATS_DRC_MEM_USAGE ]));seq_printf (m,"cache hits: %lld\n" , -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_RC_HITS ])); +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_RC_HITS ]));seq_printf (m,"cache misses: %lld\n" , -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_RC_MISSES ])); +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_RC_MISSES ]));seq_printf (m,"not cached: %lld\n" , -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_RC_NOCACHE ])); +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_RC_NOCACHE ]));seq_printf (m,"payload misses: %lld\n" ,percpu_counter_sum_positive (&nn->counter[NFSD_STATS_PAYLOAD_MISSES ]));seq_printf (m,"longest chain len: %u\n" , nn->longest_chain);
fs/nfsd/nfsctl.c +3 -9
@@ -1671,7 +1671,7 @@ static __net_init int nfsd_net_init(struct net *net) retval =nfsd_idmap_init (net);if (retval) goto out_idmap_error; - retval =nfsd_net_reply_cache_init (nn); + retval =nfsd_stat_counters_init (nn);if (retval) goto out_repcache_error; nn->nfsd_versions =NULL ;
@@ -1701,7 +1701,7 @@ static __net_exit void nfsd_net_exit(struct net *net)struct nfsd_net *nn =net_generic (net, nfsd_net_id);nfsd_proc_stat_shutdown (net); -nfsd_net_reply_cache_destroy (nn); +nfsd_stat_counters_destroy (nn);nfsd_idmap_shutdown (net);nfsd_export_shutdown (net);nfsd_netns_free_versions (nn);
@@ -1724,12 +1724,9 @@ static int __init init_nfsd(void) retval =nfsd4_init_pnfs ();if (retval) goto out_free_slabs; - retval =nfsd_stat_counters_init ();/* Statistics */ -if (retval) - goto out_free_pnfs; retval =nfsd_drc_slab_create ();if (retval) - goto out_free_stat; + goto out_free_pnfs;nfsd_lockd_init ();/* lockd->nfsd callbacks */ retval =create_proc_exports_entry ();if (retval)
@@ -1763,8 +1760,6 @@ static int __init init_nfsd(void) out_free_lockd:nfsd_lockd_shutdown ();nfsd_drc_slab_free (); -out_free_stat: -nfsd_stat_counters_destroy (); out_free_pnfs:nfsd4_exit_pnfs (); out_free_slabs:
@@ -1782,7 +1777,6 @@ static void __exit exit_nfsd(void)nfsd_drc_slab_free ();remove_proc_entry ("fs/nfs/exports" ,NULL );remove_proc_entry ("fs/nfs" ,NULL ); -nfsd_stat_counters_destroy ();nfsd_lockd_shutdown ();nfsd4_free_slabs ();nfsd4_exit_pnfs ();
fs/nfsd/nfsfh.c +2 -1
@@ -327,6 +327,7 @@ static __be32 nfsd_set_fh_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp)__be32 fh_verify (struct svc_rqst *rqstp,struct svc_fh *fhp,umode_t type,int access) { +struct nfsd_net *nn =net_generic (SVC_NET (rqstp), nfsd_net_id);struct svc_export *exp =NULL ;struct dentry *dentry;__be32 error;
@@ -395,7 +396,7 @@ fh_verify(struct svc_rqst *rqstp, struct svc_fh *fhp, umode_t type, int access) out:trace_nfsd_fh_verify_err (rqstp, fhp, type, access, error);if (error == nfserr_stale) -nfsd_stats_fh_stale_inc (exp); +nfsd_stats_fh_stale_inc (nn, exp);return error; }
fs/nfsd/stats.c +14 -12
@@ -34,15 +34,17 @@ struct svc_stat nfsd_svcstats = {static int nfsd_show (struct seq_file *seq,void *v) { +struct net *net =pde_data (file_inode (seq->file)); +struct nfsd_net *nn =net_generic (net, nfsd_net_id);int i;seq_printf (seq,"rc %lld %lld %lld\nfh %lld 0 0 0 0\nio %lld %lld\n" , -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_RC_HITS ]), -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_RC_MISSES ]), -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_RC_NOCACHE ]), -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_FH_STALE ]), -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_IO_READ ]), -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_IO_WRITE ])); +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_RC_HITS ]), +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_RC_MISSES ]), +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_RC_NOCACHE ]), +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_FH_STALE ]), +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_IO_READ ]), +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_IO_WRITE ]));/* thread usage: */ seq_printf (seq,"th %u 0" ,atomic_read (&nfsdstats.th_cnt));
@@ -63,10 +65,10 @@ static int nfsd_show(struct seq_file *seq, void *v)seq_printf (seq,"proc4ops %u" ,LAST_NFS4_OP +1 );for (i =0 ; i <=LAST_NFS4_OP ; i++) {seq_printf (seq," %lld" , -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_NFS4_OP (i)])); +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_NFS4_OP (i)])); }seq_printf (seq,"\nwdeleg_getattr %lld" , -percpu_counter_sum_positive (&nfsdstats.counter[NFSD_STATS_WDELEG_GETATTR ])); +percpu_counter_sum_positive (&nn->counter[NFSD_STATS_WDELEG_GETATTR ]));seq_putc (seq,'\n' );#endif
@@ -108,14 +110,14 @@ void nfsd_percpu_counters_destroy(struct percpu_counter counters[], int num)percpu_counter_destroy (&counters[i]); } -int nfsd_stat_counters_init (void ) +int nfsd_stat_counters_init (struct nfsd_net *nn) { -return nfsd_percpu_counters_init (nfsdstats.counter,NFSD_STATS_COUNTERS_NUM ); +return nfsd_percpu_counters_init (nn->counter,NFSD_STATS_COUNTERS_NUM ); } -void nfsd_stat_counters_destroy (void ) +void nfsd_stat_counters_destroy (struct nfsd_net *nn) { -nfsd_percpu_counters_destroy (nfsdstats.counter,NFSD_STATS_COUNTERS_NUM ); +nfsd_percpu_counters_destroy (nn->counter,NFSD_STATS_COUNTERS_NUM ); }void nfsd_proc_stat_init (struct net *net)
fs/nfsd/stats.h +19 -35
97 lines not on this page · raw message
fs/nfsd/vfs.c +3 -2
@@ -1002,7 +1002,8 @@ static __be32 nfsd_finish_read(struct svc_rqst *rqstp, struct svc_fh *fhp, unsigned long *count, u32 *eof, ssize_t host_err) { if (host_err >= 0) { - nfsd_stats_io_read_add(fhp->fh_export, host_err); + struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id); +nfsd_stats_io_read_add (nn, fhp->fh_export, host_err); *eof =nfsd_eof_on_read (file, offset, host_err, *count); *count = host_err;fsnotify_access (file);
@@ -1185,7 +1186,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf, goto out_nfserr; } *cnt = host_err; -nfsd_stats_io_write_add (exp, *cnt); +nfsd_stats_io_write_add (nn, exp, *cnt);fsnotify_modify (file); host_err =filemap_check_wb_err (file->f_mapping, since);if (host_err <0 )
Signature · 1 line
2.43.0
PATCH v2 10/13 nfsd: move th_cnt into nfsd_net
In reply to Josef BacikThis is the last global stat, move it into nfsd_net and adjust all the users to use that variant instead of the global one.
--- fs/nfsd/netns.h | 3 +++ fs/nfsd/nfssvc.c | 4 ++-- fs/nfsd/stats.c | 4 ++-- fs/nfsd/stats.h | 6 ------ 4 files changed, 7 insertions(+), 10 deletions(-)
fs/nfsd/netns.h +3 -0
@@ -179,6 +179,9 @@ struct nfsd_net {/* Per-netns stats counters */ struct percpu_counter counter[NFSD_STATS_COUNTERS_NUM ]; +/* number of available threads */ +atomic_t th_cnt; +/* longest hash chain seen */ unsigned int longest_chain;
fs/nfsd/nfssvc.c +2 -2
@@ -924,7 +924,7 @@ nfsd(void *vrqstp) current->fs->umask =0 ; -atomic_inc (&nfsdstats.th_cnt); +atomic_inc (&nn->th_cnt);set_freezable ();
@@ -940,7 +940,7 @@ nfsd(void *vrqstp)nfsd_file_net_dispose (nn); } -atomic_dec (&nfsdstats.th_cnt); +atomic_dec (&nn->th_cnt); out:/* Release the thread */
fs/nfsd/stats.c +2 -2
@@ -27,7 +27,6 @@#include "nfsd.h" -struct nfsd_stats nfsdstats;struct svc_stat nfsd_svcstats = { .program = &nfsd_program, };
@@ -47,7 +46,7 @@ static int nfsd_show(struct seq_file *seq, void *v) percpu_counter_sum_positive(&nn->counter[NFSD_STATS_IO_WRITE]));/* thread usage: */ -seq_printf (seq,"th %u 0" ,atomic_read (&nfsdstats.th_cnt)); +seq_printf (seq,"th %u 0" ,atomic_read (&nn->th_cnt));/* deprecated thread usage histogram stats */ for (i =0 ; i <10 ; i++)
@@ -112,6 +111,7 @@ void nfsd_percpu_counters_destroy(struct percpu_counter counters[], int num)int nfsd_stat_counters_init (struct nfsd_net *nn) { +atomic_set (&nn->th_cnt,0 );return nfsd_percpu_counters_init (nn->counter,NFSD_STATS_COUNTERS_NUM ); }
fs/nfsd/stats.h +0 -6
@@ -10,12 +10,6 @@#include <uapi/linux/nfsd/stats.h> #include <linux/percpu_counter.h> -struct nfsd_stats { -atomic_t th_cnt;/* number of available threads */ -}; - -extern struct nfsd_stats nfsdstats; -extern struct svc_stat nfsd_svcstats;int nfsd_percpu_counters_init (struct percpu_counter *counters,int num);
Signature · 1 line
2.43.0
PATCH v2 11/13 nfsd: make svc_stat per-network namespace instead of global
In reply to Josef BacikThe final bit of stats that is global is the rpc svc_stat. Move this into the nfsd_net struct and use that everywhere instead of the global struct. Remove the unused global struct.
--- fs/nfsd/netns.h | 4 ++++ fs/nfsd/nfsctl.c | 2 ++ fs/nfsd/nfssvc.c | 2 +- fs/nfsd/stats.c | 10 ++++------ fs/nfsd/stats.h | 2 -- 5 files changed, 11 insertions(+), 9 deletions(-)
fs/nfsd/netns.h +4 -0
@@ -14,6 +14,7 @@#include <linux/nfs4.h> #include <linux/percpu_counter.h> #include <linux/siphash.h> +#include <linux/sunrpc/stats.h> /* Hash tables for nfs4_clientid state */ #define CLIENT_HASH_BITS 4
@@ -182,6 +183,9 @@ struct nfsd_net {/* number of available threads */ atomic_t th_cnt; +/* sunrpc svc stats */ +struct svc_stat nfsd_svcstats; +/* longest hash chain seen */ unsigned int longest_chain;
fs/nfsd/nfsctl.c +2 -0
@@ -1674,6 +1674,8 @@ static __net_init int nfsd_net_init(struct net *net) retval =nfsd_stat_counters_init (nn);if (retval) goto out_repcache_error; +memset (&nn->nfsd_svcstats,0 ,sizeof (nn->nfsd_svcstats)); + nn->nfsd_svcstats.program = &nfsd_program; nn->nfsd_versions =NULL ; nn->nfsd4_minorversions =NULL ;nfsd4_init_leases_net (nn);
fs/nfsd/nfssvc.c +1 -1
@@ -660,7 +660,7 @@ int nfsd_create_serv(struct net *net)if (nfsd_max_blksize ==0 ) nfsd_max_blksize =nfsd_get_default_max_blksize ();nfsd_reset_versions (nn); - serv =svc_create_pooled (&nfsd_program, &nfsd_svcstats, + serv =svc_create_pooled (&nfsd_program, &nn->nfsd_svcstats, nfsd_max_blksize, nfsd);if (serv ==NULL )return -ENOMEM ;
fs/nfsd/stats.c +4 -6
@@ -27,10 +27,6 @@#include "nfsd.h" -struct svc_stat nfsd_svcstats = { - .program = &nfsd_program, -}; -static int nfsd_show (struct seq_file *seq,void *v) {struct net *net =pde_data (file_inode (seq->file));
@@ -56,7 +52,7 @@ static int nfsd_show(struct seq_file *seq, void *v)seq_puts (seq,"\nra 0 0 0 0 0 0 0 0 0 0 0 0\n" );/* show my rpc info */ -svc_seq_show (seq, &nfsd_svcstats); +svc_seq_show (seq, &nn->nfsd_svcstats);#ifdef CONFIG_NFSD_V4 /* Show count for individual nfsv4 operations */
@@ -122,7 +118,9 @@ void nfsd_stat_counters_destroy(struct nfsd_net *nn)void nfsd_proc_stat_init (struct net *net) { -svc_proc_register (net, &nfsd_svcstats, &nfsd_proc_ops); +struct nfsd_net *nn =net_generic (net, nfsd_net_id); + +svc_proc_register (net, &nn->nfsd_svcstats, &nfsd_proc_ops); }void nfsd_proc_stat_shutdown (struct net *net)
fs/nfsd/stats.h +0 -2
@@ -10,8 +10,6 @@#include <uapi/linux/nfsd/stats.h> #include <linux/percpu_counter.h> -extern struct svc_stat nfsd_svcstats; -int nfsd_percpu_counters_init (struct percpu_counter *counters,int num);void nfsd_percpu_counters_reset (struct percpu_counter *counters,int num);void nfsd_percpu_counters_destroy (struct percpu_counter *counters,int num);
Signature · 1 line
2.43.0
PATCH v2 12/13 nfs: expose /proc/net/sunrpc/nfs in net namespaces
In reply to Josef BacikWe're using nfs mounts inside of containers in production and noticed that the nfs stats are not exposed in /proc. This is a problem for us as we use these stats for monitoring, and have to do this awkward bind mount from the main host into the container in order to get to these states.
Add the rpc_proc_register call to the pernet operations entry and exit points so these stats can be exposed inside of network namespaces.
--- fs/nfs/inode.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-)
fs/nfs/inode.c +2 -4
@@ -2427,11 +2427,13 @@ EXPORT_SYMBOL_GPL(nfs_net_id);static int nfs_net_init (struct net *net) {nfs_clients_init (net); +rpc_proc_register (net, &nfs_rpcstat);return nfs_fs_proc_net_init (net); }static void nfs_net_exit (struct net *net) { +rpc_proc_unregister (net,"nfs" );nfs_fs_proc_net_exit (net);nfs_clients_exit (net); }
@@ -2486,15 +2488,12 @@ static int __init init_nfs_fs(void)if (err) goto out1; -rpc_proc_register (&init_net, &nfs_rpcstat); - err =register_nfs_fs ();if (err) goto out0;return 0 ; out0: -rpc_proc_unregister (&init_net,"nfs" );nfs_destroy_directcache (); out1:nfs_destroy_writepagecache ();
@@ -2524,7 +2523,6 @@ static void __exit exit_nfs_fs(void)nfs_destroy_inodecache ();nfs_destroy_nfspagecache ();unregister_pernet_subsys (&nfs_net_ops); -rpc_proc_unregister (&init_net,"nfs" );unregister_nfs_fs ();nfs_fs_proc_exit ();nfsiod_stop ();
Signature · 1 line
2.43.0
PATCH v2 13/13 nfs: make the rpc_stat per net namespace
In reply to Josef BacikNow that we're exposing the rpc stats on a per-network namespace basis, move this struct into struct nfs_net and use that to make sure only the per-network namespace stats are exposed.
--- fs/nfs/client.c | 5 ++++- fs/nfs/inode.c | 4 +++- fs/nfs/internal.h | 2 -- fs/nfs/netns.h | 2 ++ 4 files changed, 9 insertions(+), 4 deletions(-)
fs/nfs/client.c +4 -1
@@ -73,7 +73,6 @@ const struct rpc_program nfs_program = { .number = NFS_PROGRAM, .nrvers = ARRAY_SIZE(nfs_version), .version = nfs_version, - .stats = &nfs_rpcstat, .pipe_dir_name = NFS_PIPE_DIRNAME, };
@@ -502,6 +501,7 @@ int nfs_create_rpc_client(struct nfs_client *clp,const struct nfs_client_initdata *cl_init, rpc_authflavor_t flavor) { + struct nfs_net *nn = net_generic(clp->cl_net, nfs_net_id);struct rpc_clnt *clnt =NULL ; struct rpc_create_args args = { .net = clp->cl_net,
@@ -513,6 +513,7 @@ int nfs_create_rpc_client(struct nfs_client *clp, .servername = clp->cl_hostname, .nodename = cl_init->nodename, .program = &nfs_program, + .stats = &nn->rpcstats, .version = clp->rpc_ops->version, .authflavor = flavor, .cred = cl_init->cred,
@@ -1175,6 +1176,8 @@ void nfs_clients_init(struct net *net)#endif spin_lock_init (&nn->nfs_client_lock); nn->boot_time =ktime_get_real (); +memset (&nn->rpcstats,0 ,sizeof (nn->rpcstats)); + nn->rpcstats.program = &nfs_program;nfs_netns_sysfs_setup (nn, net); }
fs/nfs/inode.c +3 -1
@@ -2426,8 +2426,10 @@ EXPORT_SYMBOL_GPL(nfs_net_id);static int nfs_net_init (struct net *net) { +struct nfs_net *nn =net_generic (net, nfs_net_id); +nfs_clients_init (net); -rpc_proc_register (net, &nfs_rpcstat); +rpc_proc_register (net, &nn->rpcstats);return nfs_fs_proc_net_init (net); }
fs/nfs/internal.h +0 -2
@@ -449,8 +449,6 @@ int nfs_try_get_tree(struct fs_context *);int nfs_get_tree_common (struct fs_context *);void nfs_kill_super (struct super_block *); -extern struct rpc_stat nfs_rpcstat; -extern int __init register_nfs_fs (void ); extern void __exit unregister_nfs_fs(void);extern bool nfs_sb_active (struct super_block *sb);
fs/nfs/netns.h +2 -0
@@ -9,6 +9,7 @@#include <linux/nfs4.h> #include <net/net_namespace.h> #include <net/netns/generic.h> +#include <linux/sunrpc/stats.h> struct bl_dev_msg {int32_t status;
@@ -34,6 +35,7 @@ struct nfs_net {struct nfs_netns_client *nfs_client;spinlock_t nfs_client_lock;ktime_t boot_time; +struct rpc_stat rpcstats;#ifdef CONFIG_PROC_FS struct proc_dir_entry *proc_nfsfs;#endif
Signature · 1 line
2.43.0
Re: [PATCH v2 05/13] sunrpc: add a struct rpc_stats arg to rpc_create_args
In reply to Josef BacikQuoted 10 lines · Josef Bacik
We want to be able to have our rpc stats handled in a per network namespace manner, so add an option to
rpc_create_argsto specify a differentrpc_statsstruct instead of using the one on therpc_program.Signed-off-by Josef Bacik <[email protected]>--- fs/nfs/client.c | 2 +- include/linux/sunrpc/clnt.h | 1 + net/sunrpc/clnt.c | 2 +- 3 files changed, 3 insertions(+), 2 deletions(-)
I know it isn't obvious to an outside observer, but the maintainership of the NFS server and client are separate.
NFS client patches go To: Trond and Anna, Cc: linux-nfs
NFS server patches go To: Jeff and Chuck, Cc: linux-nfs
and you can Cc: server patches on the reviewers listed in MAINTAINERS too if you like.
Quoted 42 lines
fs/nfs/client.c +1
-1@@ -73,7 +73,7 @@ const struct rpc_program nfs_program = { .number = NFS_PROGRAM, .nrvers = ARRAY_SIZE(nfs_version), .version = nfs_version, - .stats = &nfs_rpcstat, + .stats = &nfs_rpcstat, .pipe_dir_name = NFS_PIPE_DIRNAME, };include/linux/sunrpc/clnt.h +1
-0@@ -139,6 +139,7 @@ struct rpc_create_args {const char *servername;const char *nodename;const struct rpc_program *program; +struct rpc_stat *stats;u32 prognumber;/* overrides program->number */ u32 version;rpc_authflavor_t authflavor;net/sunrpc/clnt.c +1
-1@@ -405,7 +405,7 @@ static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args, clnt->cl_maxproc = version->nrprocs; clnt->cl_prog = args->prognumber ? : program->number; clnt->cl_vers = version->number; - clnt->cl_stats = program->stats; + clnt->cl_stats = args->stats ? : program->stats; clnt->cl_metrics =rpc_alloc_iostats (clnt);rpc_init_pipe_dir_head (&clnt->cl_pipedir_objects); err = -ENOMEM ;Signature · 1 line
2.43.0
Signature · 1 line
Chuck Lever
Re: [PATCH v2 03/13] sunrpc: pass in the sv_stats struct through svc_create*
In reply to Josef BacikQuoted 106 lines · Josef Bacik
Since only one service actually reports the rpc stats there's not much of a reason to have a pointer to it in the
svc_programstruct. Adjust the svc_create* functions to take thesv_statsas an argument and pass the struct through there as desired instead of getting it from thesvc_program->pg_stats.Signed-off-by Josef Bacik <[email protected]>--- fs/lockd/svc.c | 2 +- fs/nfs/callback.c | 2 +- fs/nfsd/nfssvc.c | 3 ++- include/linux/sunrpc/svc.h | 8 ++++---- net/sunrpc/svc.c | 17 ++++++++++------- 5 files changed, 18 insertions(+), 14 deletions(-)fs/lockd/svc.c +1
-1@@ -337,7 +337,7 @@ static int lockd_get(void) nlm_timeout =LOCKD_DFLT_TIMEO ; nlmsvc_timeout = nlm_timeout *HZ ; - serv =svc_create (&nlmsvc_program,LOCKD_BUFSIZE , lockd); + serv =svc_create (&nlmsvc_program,NULL ,LOCKD_BUFSIZE , lockd);if (!serv) {printk (KERN_WARNING "lockd_up: create service failed\n" );return -ENOMEM ;fs/nfs/callback.c +1
-1@@ -202,7 +202,7 @@ static struct svc_serv *nfs_callback_create_svc(int minorversion)if (minorversion)return ERR_PTR (-ENOTSUPP );#endif - serv =svc_create (&nfs4_callback_program,NFS4_CALLBACK_BUFSIZE , + serv =svc_create (&nfs4_callback_program,NULL ,NFS4_CALLBACK_BUFSIZE , threadfn);if (!serv) {printk (KERN_ERR "nfs_callback_create_svc: create service failed\n" );fs/nfsd/nfssvc.c +2
-1@@ -661,7 +661,8 @@ int nfsd_create_serv(struct net *net)if (nfsd_max_blksize ==0 ) nfsd_max_blksize =nfsd_get_default_max_blksize ();nfsd_reset_versions (nn); - serv =svc_create_pooled (&nfsd_program, nfsd_max_blksize, nfsd); + serv =svc_create_pooled (&nfsd_program, &nfsd_svcstats, + nfsd_max_blksize, nfsd);if (serv ==NULL )return -ENOMEM ;include/linux/sunrpc/svc.h +4
-4@@ -402,8 +402,8 @@ struct svc_procedure {int svc_rpcb_setup (struct svc_serv *serv,struct net *net);void svc_rpcb_cleanup (struct svc_serv *serv,struct net *net);int svc_bind (struct svc_serv *serv,struct net *net); -struct svc_serv *svc_create (struct svc_program *,unsigned int , -int (*threadfn)(void *data)); +struct svc_serv *svc_create (struct svc_program *,struct svc_stat *, +unsigned int ,int (*threadfn)(void *data));struct svc_rqst *svc_rqst_alloc (struct svc_serv *serv,struct svc_pool *pool,int node); bool svc_rqst_replace_page(struct svc_rqst *rqstp,@@ -411,8 +411,8 @@ bool svc_rqst_replace_page(struct svc_rqst *rqstp,void svc_rqst_release_pages (struct svc_rqst *rqstp);void svc_rqst_free (struct svc_rqst *);void svc_exit_thread (struct svc_rqst *); -struct svc_serv *svc_create_pooled (struct svc_program *,unsigned int , -int (*threadfn)(void *data)); +struct svc_serv *svc_create_pooled (struct svc_program *,struct svc_stat *, +unsigned int ,int (*threadfn)(void *data));int svc_set_num_threads (struct svc_serv *,struct svc_pool *,int );int svc_pool_stats_open (struct svc_info *si,struct file *file);void svc_process (struct svc_rqst *rqstp);net/sunrpc/svc.c +3
-3@@ -451,8 +451,8 @@ __svc_init_bc(struct svc_serv *serv)* Create an RPC service */ static struct svc_serv * -__svc_create (struct svc_program *prog,unsigned int bufsize,int npools, -int (*threadfn)(void *data)) +__svc_create (struct svc_program *prog,struct svc_stat *stats, +unsigned int bufsize,int npools,int (*threadfn)(void *data)) {struct svc_serv *serv;unsigned int vers;@@ -463,7 +463,7 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,return NULL ; serv->sv_name = prog->pg_name; serv->sv_program = prog; - serv->sv_stats = prog->pg_stats; + serv->sv_stats = stats;if (bufsize >RPCSVC_MAXPAYLOAD ) bufsize =RPCSVC_MAXPAYLOAD ; serv->sv_max_payload = bufsize? bufsize :4096 ;
net/sunrpc/svc.c __svc_create(struct svc_program *prog, unsigned int bufsize, int npools, · line 529
Quoted 9 lines
net/sunrpc/svc.c +1
-0@@ -521,34 +521,37 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,/** * svc_create - Create an RPC service * @prog: the RPC program the new service will handle +* @stats: the stats struct if desired * @bufsize: maximum message size for @prog * @threadfn: a function to service RPC requests for @prog * * Returns an instantiated struct svc_serv object or NULL. */
Here's the only minor quibble I have so far:
svc_create's callers don't use stats, so maybe you don't need to add an @stats argument for this API.
Fwiw, I haven't gotten all the way to the end of the series yet.
Quoted 36 lines
net/sunrpc/svc.c +6
-4-struct svc_serv *svc_create (struct svc_program *prog,unsigned int bufsize, -int (*threadfn)(void *data)) +struct svc_serv *svc_create (struct svc_program *prog,struct svc_stat *stats, +unsigned int bufsize,int (*threadfn)(void *data)) { -return __svc_create (prog, bufsize,1 , threadfn); +return __svc_create (prog, stats, bufsize,1 , threadfn); }EXPORT_SYMBOL_GPL (svc_create);/** * svc_create_pooled - Create an RPC service with pooled threads * @prog: the RPC program the new service will handle +* @stats: the stats struct if desired * @bufsize: maximum message size for @prog * @threadfn: a function to service RPC requests for @prog * * Returns an instantiated struct svc_serv object or NULL. */ struct svc_serv *svc_create_pooled (struct svc_program *prog, +struct svc_stat *stats,unsigned int bufsize,int (*threadfn)(void *data)) {struct svc_serv *serv;unsigned int npools =svc_pool_map_get (); - serv =__svc_create (prog, bufsize, npools, threadfn); + serv =__svc_create (prog, stats, bufsize, npools, threadfn);if (!serv) goto out_err;return serv;Signature · 1 line
2.43.0
Signature · 1 line
Chuck Lever
Re: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
In reply to Josef BacikQuoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.
Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
Quoted 94 lines
Signed-off-by Josef Bacik <[email protected]>--- fs/nfsd/netns.h | 3 +++ fs/nfsd/nfssvc.c | 4 ++-- fs/nfsd/stats.c | 4 ++-- fs/nfsd/stats.h | 6 ------ 4 files changed, 7 insertions(+), 10 deletions(-)fs/nfsd/netns.h +3
-0@@ -179,6 +179,9 @@ struct nfsd_net {/* Per-netns stats counters */ struct percpu_counter counter[NFSD_STATS_COUNTERS_NUM ]; +/* number of available threads */ +atomic_t th_cnt; +/* longest hash chain seen */ unsigned int longest_chain;fs/nfsd/nfssvc.c +2
-2@@ -924,7 +924,7 @@ nfsd(void *vrqstp) current->fs->umask =0 ; -atomic_inc (&nfsdstats.th_cnt); +atomic_inc (&nn->th_cnt);set_freezable ();@@ -940,7 +940,7 @@ nfsd(void *vrqstp)nfsd_file_net_dispose (nn); } -atomic_dec (&nfsdstats.th_cnt); +atomic_dec (&nn->th_cnt); out:/* Release the thread */ fs/nfsd/stats.c +2
-2@@ -27,7 +27,6 @@#include "nfsd.h" -struct nfsd_stats nfsdstats;struct svc_stat nfsd_svcstats = { .program = &nfsd_program, };@@ -47,7 +46,7 @@ static int nfsd_show(struct seq_file *seq, void *v) percpu_counter_sum_positive(&nn->counter[NFSD_STATS_IO_WRITE]));/* thread usage: */ -seq_printf (seq,"th %u 0" ,atomic_read (&nfsdstats.th_cnt)); +seq_printf (seq,"th %u 0" ,atomic_read (&nn->th_cnt));/* deprecated thread usage histogram stats */ for (i =0 ; i <10 ; i++)@@ -112,6 +111,7 @@ void nfsd_percpu_counters_destroy(struct percpu_counter counters[], int num)int nfsd_stat_counters_init (struct nfsd_net *nn) { +atomic_set (&nn->th_cnt,0 );return nfsd_percpu_counters_init (nn->counter,NFSD_STATS_COUNTERS_NUM ); }fs/nfsd/stats.h +0
-6@@ -10,12 +10,6 @@#include <uapi/linux/nfsd/stats.h> #include <linux/percpu_counter.h> -struct nfsd_stats { -atomic_t th_cnt;/* number of available threads */ -}; - -extern struct nfsd_stats nfsdstats; -extern struct svc_stat nfsd_svcstats;int nfsd_percpu_counters_init (struct percpu_counter *counters,int num);Signature · 1 line
2.43.0
Signature · 1 line
Chuck Lever
Re: [PATCH v2 05/13] sunrpc: add a struct rpc_stats arg to rpc_create_args
In reply to Chuck LeverQuoted 15 lines · Chuck Lever
Quoted 10 lines · Josef Bacik
We want to be able to have our rpc stats handled in a per network namespace manner, so add an option to
rpc_create_argsto specify a differentrpc_statsstruct instead of using the one on therpc_program.Signed-off-by Josef Bacik <[email protected]>--- fs/nfs/client.c | 2 +- include/linux/sunrpc/clnt.h | 1 + net/sunrpc/clnt.c | 2 +- 3 files changed, 3 insertions(+), 2 deletions(-)I know it isn't obvious to an outside observer, but the maintainership of the NFS server and client are separate.
Quoted 6 lines
NFS client patches go To: Trond and Anna, Cc: linux-nfs
NFS server patches go To: Jeff and Chuck, Cc: linux-nfs
and you can Cc: server patches on the reviewers listed in
MAINTAINERStoo if you like.
Sounds good, should I split the series up completely? I think the nfsd and nfs sunrpc related changes are unique to either stack so I can split it out into two different series if that would help. Thanks,
Josef
Re: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
In reply to Chuck LeverQuoted 9 lines · Chuck Lever
Quoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
I don't think we want the network namespaces seeing how many threads exist in the entire system right?
Additionally it appears that we can have multiple threads per network namespace, so it's not like this will just show 1 for each individual nn, it'll show however many threads have been configured for that nfsd in that network namespace.
I'm good either way, but it makes sense to me to only surface the network namespace related thread count. I could probably have a global counter and only surface the global counter if net == &init_net. Let me know what you prefer. Thanks,
Josef
Re: [PATCH v2 03/13] sunrpc: pass in the sv_stats struct through svc_create*
In reply to Chuck LeverQuoted 119 lines · Chuck Lever
Quoted 106 lines · Josef Bacik
Since only one service actually reports the rpc stats there's not much of a reason to have a pointer to it in the
svc_programstruct. Adjust the svc_create* functions to take thesv_statsas an argument and pass the struct through there as desired instead of getting it from thesvc_program->pg_stats.Signed-off-by Josef Bacik <[email protected]>--- fs/lockd/svc.c | 2 +- fs/nfs/callback.c | 2 +- fs/nfsd/nfssvc.c | 3 ++- include/linux/sunrpc/svc.h | 8 ++++---- net/sunrpc/svc.c | 17 ++++++++++------- 5 files changed, 18 insertions(+), 14 deletions(-)fs/lockd/svc.c +1
-1@@ -337,7 +337,7 @@ static int lockd_get(void) nlm_timeout =LOCKD_DFLT_TIMEO ; nlmsvc_timeout = nlm_timeout *HZ ; - serv =svc_create (&nlmsvc_program,LOCKD_BUFSIZE , lockd); + serv =svc_create (&nlmsvc_program,NULL ,LOCKD_BUFSIZE , lockd);if (!serv) {printk (KERN_WARNING "lockd_up: create service failed\n" );return -ENOMEM ;fs/nfs/callback.c +1
-1@@ -202,7 +202,7 @@ static struct svc_serv *nfs_callback_create_svc(int minorversion)if (minorversion)return ERR_PTR (-ENOTSUPP );#endif - serv =svc_create (&nfs4_callback_program,NFS4_CALLBACK_BUFSIZE , + serv =svc_create (&nfs4_callback_program,NULL ,NFS4_CALLBACK_BUFSIZE , threadfn);if (!serv) {printk (KERN_ERR "nfs_callback_create_svc: create service failed\n" );fs/nfsd/nfssvc.c +2
-1@@ -661,7 +661,8 @@ int nfsd_create_serv(struct net *net)if (nfsd_max_blksize ==0 ) nfsd_max_blksize =nfsd_get_default_max_blksize ();nfsd_reset_versions (nn); - serv =svc_create_pooled (&nfsd_program, nfsd_max_blksize, nfsd); + serv =svc_create_pooled (&nfsd_program, &nfsd_svcstats, + nfsd_max_blksize, nfsd);if (serv ==NULL )return -ENOMEM ;include/linux/sunrpc/svc.h +4
-4@@ -402,8 +402,8 @@ struct svc_procedure {int svc_rpcb_setup (struct svc_serv *serv,struct net *net);void svc_rpcb_cleanup (struct svc_serv *serv,struct net *net);int svc_bind (struct svc_serv *serv,struct net *net); -struct svc_serv *svc_create (struct svc_program *,unsigned int , -int (*threadfn)(void *data)); +struct svc_serv *svc_create (struct svc_program *,struct svc_stat *, +unsigned int ,int (*threadfn)(void *data));struct svc_rqst *svc_rqst_alloc (struct svc_serv *serv,struct svc_pool *pool,int node); bool svc_rqst_replace_page(struct svc_rqst *rqstp,@@ -411,8 +411,8 @@ bool svc_rqst_replace_page(struct svc_rqst *rqstp,void svc_rqst_release_pages (struct svc_rqst *rqstp);void svc_rqst_free (struct svc_rqst *);void svc_exit_thread (struct svc_rqst *); -struct svc_serv *svc_create_pooled (struct svc_program *,unsigned int , -int (*threadfn)(void *data)); +struct svc_serv *svc_create_pooled (struct svc_program *,struct svc_stat *, +unsigned int ,int (*threadfn)(void *data));int svc_set_num_threads (struct svc_serv *,struct svc_pool *,int );int svc_pool_stats_open (struct svc_info *si,struct file *file);void svc_process (struct svc_rqst *rqstp);net/sunrpc/svc.c +3
-3@@ -451,8 +451,8 @@ __svc_init_bc(struct svc_serv *serv)* Create an RPC service */ static struct svc_serv * -__svc_create (struct svc_program *prog,unsigned int bufsize,int npools, -int (*threadfn)(void *data)) +__svc_create (struct svc_program *prog,struct svc_stat *stats, +unsigned int bufsize,int npools,int (*threadfn)(void *data)) {struct svc_serv *serv;unsigned int vers;@@ -463,7 +463,7 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,return NULL ; serv->sv_name = prog->pg_name; serv->sv_program = prog; - serv->sv_stats = prog->pg_stats; + serv->sv_stats = stats;if (bufsize >RPCSVC_MAXPAYLOAD ) bufsize =RPCSVC_MAXPAYLOAD ; serv->sv_max_payload = bufsize? bufsize :4096 ;net/sunrpc/svc.c __svc_create(struct svc_program *prog, unsigned int bufsize, int npools, · line 529
Quoted 9 lines
net/sunrpc/svc.c +1
-0@@ -521,34 +521,37 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,/** * svc_create - Create an RPC service * @prog: the RPC program the new service will handle +* @stats: the stats struct if desired * @bufsize: maximum message size for @prog * @threadfn: a function to service RPC requests for @prog * * Returns an instantiated struct svc_serv object or NULL. */ Here's the only minor quibble I have so far:
Quoted 4 lines
svc_create's callers don't use stats, so maybe you don't need to add an @stats argument for this API.
Fwiw, I haven't gotten all the way to the end of the series yet.
Yup you're right, I can drop this bit. Thanks,
Josef
Jeff Layton
maintainerRe: [PATCH v2 05/13] sunrpc: add a struct rpc_stats arg to rpc_create_args
In reply to Josef BacikQuoted 18 lines · Josef Bacik
Quoted 17 lines · Chuck Lever
Quoted 10 lines · Josef Bacik
We want to be able to have our rpc stats handled in a per network namespace manner, so add an option to
rpc_create_argsto specify a differentrpc_statsstruct instead of using the one on therpc_program.Signed-off-by Josef Bacik <[email protected]>--- fs/nfs/client.c | 2 +- include/linux/sunrpc/clnt.h | 1 + net/sunrpc/clnt.c | 2 +- 3 files changed, 3 insertions(+), 2 deletions(-)I know it isn't obvious to an outside observer, but the maintainership of the NFS server and client are separate.
NFS client patches go To: Trond and Anna, Cc: linux-nfs
Quoted 9 lines
Quoted 4 lines
NFS server patches go To: Jeff and Chuck, Cc: linux-nfs
and you can Cc: server patches on the reviewers listed in
MAINTAINERStoo if you like.Sounds good, should I split the series up completely? I think the nfsd and nfs sunrpc related changes are unique to either stack so I can split it out into two different series if that would help. Thanks,
Yes, that'd probably be best. The client-side changes will probably go in via Trond or Anna, and Chuck will (likely) take the server-side changes.
Cheers,
Signature · 1 line
Jeff Layton <[email protected]>
Jeff Layton
maintainerRe: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
In reply to Josef BacikQuoted 11 lines · Josef Bacik
Quoted 9 lines · Chuck Lever
Quoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
Each container is going to start /proc/fs/nfsd/threads number of threads regardless. I hadn't actually grokked that they just get tossed onto the pile of threads that service requests.
Is is possible for one container to start a small number of threads but have its client load be such that it spills over and ends up stealing threads from other containers?
Quoted 8 lines
I don't think we want the network namespaces seeing how many threads exist in the entire system right?
Additionally it appears that we can have multiple threads per network namespace, so it's not like this will just show 1 for each individual nn, it'll show however many threads have been configured for that nfsd in that network namespace.
Quoted 5 lines
I'm good either way, but it makes sense to me to only surface the network namespace related thread count. I could probably have a global counter and only surface the global counter if net ==
&init_net. Let me know what you prefer. Thanks,
The old stat meant "number of threads that are currently busy". The new one will mean "number of threads that are busy servicing requests in this namespace". The latter seems more useful on a per-ns basis.
In fact, it might be interesting to throw out some kind of warning when the th_cnt exceeds the number of threads that you've allocated in the container. That might be an indicator that you didn't spin up enough and are poaching from other containers.
Or... maybe we should be queueing the RPC when that happens so that you can't use more than you've allocated for the container?
Signature · 1 line
Jeff Layton <[email protected]>
Quoted 56 lines · Josef Bacik
v1: https://lore.kernel.org/linux-nfs/[email protected]/
v1->v2:
- rework the sunprc service creation to take a pointer to thesv_stats.
- dropped ->pg_stats from thesvc_program.
- converted all of the nfsd global stats to per-network namespace.
- added the ability to point at a specificrpc_statfor rpc program creation.
- converted the rpc stats for nfs to per-network namespace.-- Original email --
Hello,We're currently deploying NFS internally and have run into some oddities with our usage of containers. All of the services that mount and export NFS volumes run inside of containers, specifically all the namespaces including network namespaces. Our monitoring is done on a per-container basis, so we need access to the nfs and nfsd stats that are under
/proc/net/sunrpc. However these are only tied to theinit_net, which makes them invisible to containers in a different network namespace.Fix this so that these files are tied to the network namespace. This allows us to avoid the hack of bind mounting the hosts /proc into the container in order to do proper monitoring. Thanks,
Josef
Josef Bacik (13):
sunrpc: don't change ->sv_stats if it doesn't exist
nfs: stop setting ->pg_stats for unused stats
sunrpc: pass in thesv_statsstruct through svc_create*
sunrpc: remove ->pg_stats fromsvc_program
sunrpc: add a structrpc_statsarg torpc_create_args
sunrpc: use the struct net as the svc proc private
nfsd: rename NFSD_NET_* to NFSD_STATS_*
nfsd: expose/proc/net/sunrpc/nfsdin net namespaces
nfsd: make all of the nfsd stats per-network namespace
nfsd: moveth_cntintonfsd_net
nfsd: makesvc_statper-network namespace instead of global
nfs: expose/proc/net/sunrpc/nfsin net namespaces
nfs: make therpc_statper net namespacefs/lockd/svc.c | 5 +-- fs/nfs/callback.c | 5 +-- fs/nfs/client.c | 5 ++- fs/nfs/inode.c | 8 ++--- fs/nfs/internal.h | 2 -- fs/nfs/netns.h | 2 ++ fs/nfsd/cache.h | 2 -- fs/nfsd/netns.h | 28 ++++++++++++--- fs/nfsd/nfs4proc.c | 6 ++-- fs/nfsd/nfs4state.c | 3 +- fs/nfsd/nfscache.c | 40 +++++---------------- fs/nfsd/nfsctl.c | 16 ++++----- fs/nfsd/nfsfh.c | 3 +- fs/nfsd/nfssvc.c | 13 +++---- fs/nfsd/stats.c | 53 ++++++++++++----------------
Quoted 9 lines
fs/nfsd/stats.h | 70 +++++++++++++------------------------ fs/nfsd/vfs.c | 5 +-- include/linux/sunrpc/clnt.h | 1 + include/linux/sunrpc/svc.h | 9 +++-- net/sunrpc/clnt.c | 2 +- net/sunrpc/stats.c | 2 +- net/sunrpc/svc.c | 44 ++++++++++++++--------- 22 files changed, 147 insertions(+), 177 deletions(-)
Nice work!
I took a look and this all looks good to me, modulo the nits that Chuck pointed out yesterday. I'd also plan to split this into client and server sets since they'll probably go in separately. Otherwise though, this looks good to me and you can add:
Re: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
In reply to Jeff LaytonQuoted 15 lines · Jeff Layton
Quoted 9 lines · Josef Bacik
Quoted 8 lines · Chuck Lever
Quoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
Quoted 7 lines
Each container is going to start
/proc/fs/nfsd/threadsnumber of threads regardless. I hadn't actually grokked that they just get tossed onto the pile of threads that service requests.Is is possible for one container to start a small number of threads but have its client load be such that it spills over and ends up stealing threads from other containers?
I haven't seen any code that manages resources based on namespace, except in filecache.c to restrict writeback per namespace.
My impression is that any nfsd thread can serve any namespace. I'm not sure it is currently meaningful for a particular net namespace to "create" more threads.
If someone would like that level of control, we could implement a cgroup mechanism and have one or more separate svc_pools per net namespace, maybe? </hand wave>
Quoted 2 lines
Quoted 2 lines
I don't think we want the network namespaces seeing how many threads exist in the entire system right?
If someone in a non-init net namespace does a "pgrep -c nfsd" don't they see the total nfsd thread count for the host?
Quoted 4 lines
Quoted 4 lines
Additionally it appears that we can have multiple threads per network namespace, so it's not like this will just show 1 for each individual nn, it'll show however many threads have been configured for that nfsd in that network namespace.
I've never tried this, so I'm speculating. But it seems like for now, because all nfsd threads can serve all namespaces, they should all see the global thread count stat.
Then later we can refine it.
Quoted 3 lines
Quoted 3 lines
I'm good either way, but it makes sense to me to only surface the network namespace related thread count. I could probably have a global counter and only surface the global counter if net ==
&init_net. Let me know what you prefer.
--
Chuck Lever
Re: [PATCH v2 05/13] sunrpc: add a struct rpc_stats arg to rpc_create_args
In reply to Josef BacikQuoted 22 lines · Josef Bacik
Quoted 17 lines · Chuck Lever
Quoted 10 lines · Josef Bacik
We want to be able to have our rpc stats handled in a per network namespace manner, so add an option to
rpc_create_argsto specify a differentrpc_statsstruct instead of using the one on therpc_program.Signed-off-by Josef Bacik <[email protected]>---
fs/nfs/client.c | 2 +- include/linux/sunrpc/clnt.h | 1 + net/sunrpc/clnt.c | 2 +- 3 files changed, 3 insertions(+), 2 deletions(-)I know it isn't obvious to an outside observer, but the maintainership of the NFS server and client are separate.
NFS client patches go To: Trond and Anna, Cc: linux-nfs
Quoted 7 lines
Quoted 4 lines
NFS server patches go To: Jeff and Chuck, Cc: linux-nfs
and you can Cc: server patches on the reviewers listed in
MAINTAINERStoo if you like.Sounds good, should I split the series up completely? I think the nfsd and nfs sunrpc related changes are unique to either stack
That's what it looks like to me too.
Quoted 1 line
so I can split it out into two different series if that would help.
Let's do that for v3.
--
Chuck Lever
Jeff Layton
maintainerRe: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
In reply to Chuck Lever IIIQuoted 25 lines · Chuck Lever III
Quoted 14 lines · Jeff Layton
Quoted 9 lines · Josef Bacik
Quoted 8 lines · Chuck Lever
Quoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
Quoted 7 lines
Each container is going to start
/proc/fs/nfsd/threadsnumber of threads regardless. I hadn't actually grokked that they just get tossed onto the pile of threads that service requests.Is is possible for one container to start a small number of threads but have its client load be such that it spills over and ends up stealing threads from other containers?
I haven't seen any code that manages resources based on namespace, except in
filecache.cto restrict writeback per namespace.
Quoted 8 lines
My impression is that any nfsd thread can serve any namespace. I'm not sure it is currently meaningful for a particular net namespace to "create" more threads.
If someone would like that level of control, we could implement a cgroup mechanism and have one or more separate
svc_poolsper net namespace, maybe? </hand wave>
AFAICT, the total number of threads on the system will be the sum of the threads started in each of the containers. They do just go into a big pile, and whichever one wakes up will service the request, so the threads aren't associated with the netns, per-se. The svc_rqst's however _are_ per-netns. So, I don't see anything that ensures that a container doesn't exceed the number of threads it started on its own behalf.
<hand wave>
I'm not sure we'd need to tie this in to cgroups. Now that Josef is moving some of these key structures to be per-net, it should be fairly simple to have nfsd() just look at the th_cnt and the thread count in the current namespace, and just enqueue the RPC rather than doing it? </hand wave>
OTOH, maybe I'm overly concerned here.
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
I don't think we want the network namespaces seeing how many threads exist in the entire system right?
If someone in a non-init net namespace does a "pgrep -c nfsd" don't they see the total nfsd thread count for the host?
Yes, they're kernel threads and they aren't associated with a particular pid namespace.
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines
Additionally it appears that we can have multiple threads per network namespace, so it's not like this will just show 1 for each individual nn, it'll show however many threads have been configured for that nfsd in that network namespace.
Quoted 6 lines
I've never tried this, so I'm speculating. But it seems like for now, because all nfsd threads can serve all namespaces, they should all see the global thread count stat.
Then later we can refine it.
I don't think that info is particularly useful though, and it certainly breaks expectations wrt container isolation.
Look at it this way:
Suppose I have access to a container and I spin up nfsd with a particular number of threads. I now want to know "did I spin up enough threads?" By making this per-namespace as Josef suggests it should be fairly simple to tell whether my clients are regularly overrunning the threads I started. With this info as global, I have no idea what netns the RPCs being counted are against. I can't do anything with that info.
Quoted 5 lines
Quoted 4 lines
Quoted 4 lines
I'm good either way, but it makes sense to me to only surface the network namespace related thread count. I could probably have a global counter and only surface the global counter if net ==
&init_net. Let me know what you prefer.
Signature · 1 line
Jeff Layton <[email protected]>
Re: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
Quoted 46 lines · Jeff Layton
Quoted 25 lines · Chuck Lever III
Quoted 14 lines · Jeff Layton
Quoted 9 lines · Josef Bacik
Quoted 8 lines · Chuck Lever
Quoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
Quoted 7 lines
Each container is going to start
/proc/fs/nfsd/threadsnumber of threads regardless. I hadn't actually grokked that they just get tossed onto the pile of threads that service requests.Is is possible for one container to start a small number of threads but have its client load be such that it spills over and ends up stealing threads from other containers?
I haven't seen any code that manages resources based on namespace, except in
filecache.cto restrict writeback per namespace.Quoted 8 lines
My impression is that any nfsd thread can serve any namespace. I'm not sure it is currently meaningful for a particular net namespace to "create" more threads.
If someone would like that level of control, we could implement a cgroup mechanism and have one or more separate
svc_poolsper net namespace, maybe? </hand wave>AFAICT, the total number of threads on the system will be the sum of the threads started in each of the containers. They do just go into a big pile, and whichever one wakes up will service the request, so the threads aren't associated with the netns, per-se. The svc_rqst's however _are_ per-netns. So, I don't see anything that ensures that a container doesn't exceed the number of threads it started on its own behalf.
Quoted 2 lines
<hand wave>
I'm not sure we'd need to tie this in to cgroups.
Not a need, but cgroups are typically the way that such resource constraints are managed, that's all.
Quoted 33 lines
Now that Josef is
moving some of these key structures to be per-net, it should be fairly simple to havenfsd()just look at theth_cntand the thread count in the current namespace, and just enqueue the RPC rather than doing it? </hand wave>OTOH, maybe I'm overly concerned here.
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
I don't think we want the network namespaces seeing how many threads exist in the entire system right?
If someone in a non-init net namespace does a "pgrep -c nfsd" don't they see the total nfsd thread count for the host?
Yes, they're kernel threads and they aren't associated with a particular pid namespace.
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines
Additionally it appears that we can have multiple threads per network namespace, so it's not like this will just show 1 for each individual nn, it'll show however many threads have been configured for that nfsd in that network namespace.
Quoted 6 lines
I've never tried this, so I'm speculating. But it seems like for now, because all nfsd threads can serve all namespaces, they should all see the global thread count stat.
Then later we can refine it.
Quoted 8 lines
I don't think that info is particularly useful though, and it certainly breaks expectations wrt container isolation.
Look at it this way:
Suppose I have access to a container and I spin up nfsd with a particular number of threads. I now want to know "did I spin up enough threads?"
It makes sense to me for a container to ask for one or more NFSD listeners. But I'm not yet clear on what it means for a container to try to alter the NFSD thread count, which is currently global.
Quoted 4 lines
By making this per-namespace as Josef suggests it should be fairly simple to tell whether my clients are regularly overrunning the threads I started. With this info as global, I have no idea what netns the RPCs being counted are against. I can't do anything with that info.
That's true. I'm just not sure we need to do anything significant about it as part of this patch series.
I'm not at all advocating leaving it like this. I agree it needs updating.
For now, just fill in that thread count stat with the global thread count, and later when we have a better concept for what it means to "adjust the nfsd thread count from inside a container" we can come back and make it make sense.
--
Chuck Lever
Jeff Layton
maintainerRe: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
Quoted 90 lines · Chuck Lever III
Quoted 45 lines · Jeff Layton
Quoted 25 lines · Chuck Lever III
Quoted 14 lines · Jeff Layton
Quoted 9 lines · Josef Bacik
Quoted 8 lines · Chuck Lever
Quoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
Quoted 7 lines
Each container is going to start
/proc/fs/nfsd/threadsnumber of threads regardless. I hadn't actually grokked that they just get tossed onto the pile of threads that service requests.Is is possible for one container to start a small number of threads but have its client load be such that it spills over and ends up stealing threads from other containers?
I haven't seen any code that manages resources based on namespace, except in
filecache.cto restrict writeback per namespace.Quoted 8 lines
My impression is that any nfsd thread can serve any namespace. I'm not sure it is currently meaningful for a particular net namespace to "create" more threads.
If someone would like that level of control, we could implement a cgroup mechanism and have one or more separate
svc_poolsper net namespace, maybe? </hand wave>AFAICT, the total number of threads on the system will be the sum of the threads started in each of the containers. They do just go into a big pile, and whichever one wakes up will service the request, so the threads aren't associated with the netns, per-se. The svc_rqst's however _are_ per-netns. So, I don't see anything that ensures that a container doesn't exceed the number of threads it started on its own behalf.
Quoted 2 lines
<hand wave>
I'm not sure we'd need to tie this in to cgroups.Not a need, but cgroups are typically the way that such resource constraints are managed, that's all.
Quoted 38 lines
Now that Josef is
moving some of these key structures to be per-net, it should be fairly simple to havenfsd()just look at theth_cntand the thread count in the current namespace, and just enqueue the RPC rather than doing it? </hand wave>OTOH, maybe I'm overly concerned here.
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
I don't think we want the network namespaces seeing how many threads exist in the entire system right?
If someone in a non-init net namespace does a "pgrep -c nfsd" don't they see the total nfsd thread count for the host?
Yes, they're kernel threads and they aren't associated with a particular pid namespace.
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines
Additionally it appears that we can have multiple threads per network namespace, so it's not like this will just show 1 for each individual nn, it'll show however many threads have been configured for that nfsd in that network namespace.
Quoted 6 lines
I've never tried this, so I'm speculating. But it seems like for now, because all nfsd threads can serve all namespaces, they should all see the global thread count stat.
Then later we can refine it.
I don't think that info is particularly useful though, and it certainly breaks expectations wrt container isolation.
Look at it this way:
Quoted 9 lines
Quoted 3 lines
Suppose I have access to a container and I spin up nfsd with a particular number of threads. I now want to know "did I spin up enough threads?"
It makes sense to me for a container to ask for one or more NFSD listeners. But I'm not yet clear on what it means for a container to try to alter the NFSD thread count, which is currently global.
write_threads sets the number of nfsd threads for the svc_serv, which is a per-net object, so serv->sv_nrthreads is only counting the threads for its namespace.
Each container that starts an nfsd will contribute "threads" number of nfsds to the pile. When you set "threads" to a different number, the total pile is grown or reduced accordingly. In fact, I'm not even sure we keep a systemwide total.
What _isn't_ done (unless I'm missing something) is any sort of limitation on the use of those threads. So you could have a container that starts one nfsd thread, but its clients can still have many more RPCs in flight, up to the total number of nfsd's running on the host. Limiting that might be possible, but I'm not yet convinced that it's a good idea.
It might be interesting to gather some metrics though. How often do the number of RPCs in flight exceed the number of threads that we started in a namespace? Might also be good to gather a high-water mark too?
Quoted 9 lines
Quoted 5 lines
By making this per-namespace as Josef suggests it should be fairly simple to tell whether my clients are regularly overrunning the threads I started. With this info as global, I have no idea what netns the RPCs being counted are against. I can't do anything with that info.
That's true. I'm just not sure we need to do anything significant about it as part of this patch series.
Quoted 7 lines
I'm not at all advocating leaving it like this. I agree it needs updating.
For now, just fill in that thread count stat with the global thread count, and later when we have a better concept for what it means to "adjust the nfsd thread count from inside a container" we can come back and make it make sense.
It would be good to step back and decide how we think this should work. What would this look like if we were designing it today? Then we can decide how to get there.
Personally, I think keeping the view inside the container as isolated as possible is the right thing to do.
Signature · 1 line
Jeff Layton <[email protected]>
Re: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
Quoted 105 lines · Jeff Layton
Quoted 90 lines · Chuck Lever III
Quoted 45 lines · Jeff Layton
Quoted 25 lines · Chuck Lever III
Quoted 14 lines · Jeff Layton
Quoted 9 lines · Josef Bacik
Quoted 8 lines · Chuck Lever
Quoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
Quoted 7 lines
Each container is going to start
/proc/fs/nfsd/threadsnumber of threads regardless. I hadn't actually grokked that they just get tossed onto the pile of threads that service requests.Is is possible for one container to start a small number of threads but have its client load be such that it spills over and ends up stealing threads from other containers?
I haven't seen any code that manages resources based on namespace, except in
filecache.cto restrict writeback per namespace.Quoted 8 lines
My impression is that any nfsd thread can serve any namespace. I'm not sure it is currently meaningful for a particular net namespace to "create" more threads.
If someone would like that level of control, we could implement a cgroup mechanism and have one or more separate
svc_poolsper net namespace, maybe? </hand wave>AFAICT, the total number of threads on the system will be the sum of the threads started in each of the containers. They do just go into a big pile, and whichever one wakes up will service the request, so the threads aren't associated with the netns, per-se. The svc_rqst's however _are_ per-netns. So, I don't see anything that ensures that a container doesn't exceed the number of threads it started on its own behalf.
Quoted 2 lines
<hand wave>
I'm not sure we'd need to tie this in to cgroups.Not a need, but cgroups are typically the way that such resource constraints are managed, that's all.
Quoted 38 lines
Now that Josef is
moving some of these key structures to be per-net, it should be fairly simple to havenfsd()just look at theth_cntand the thread count in the current namespace, and just enqueue the RPC rather than doing it? </hand wave>OTOH, maybe I'm overly concerned here.
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
I don't think we want the network namespaces seeing how many threads exist in the entire system right?
If someone in a non-init net namespace does a "pgrep -c nfsd" don't they see the total nfsd thread count for the host?
Yes, they're kernel threads and they aren't associated with a particular pid namespace.
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines
Additionally it appears that we can have multiple threads per network namespace, so it's not like this will just show 1 for each individual nn, it'll show however many threads have been configured for that nfsd in that network namespace.
Quoted 6 lines
I've never tried this, so I'm speculating. But it seems like for now, because all nfsd threads can serve all namespaces, they should all see the global thread count stat.
Then later we can refine it.
I don't think that info is particularly useful though, and it certainly breaks expectations wrt container isolation.
Look at it this way:
Quoted 9 lines
Quoted 3 lines
Suppose I have access to a container and I spin up nfsd with a particular number of threads. I now want to know "did I spin up enough threads?"
It makes sense to me for a container to ask for one or more NFSD listeners. But I'm not yet clear on what it means for a container to try to alter the NFSD thread count, which is currently global.
Quoted 3 lines
write_threadssets the number of nfsd threads for thesvc_serv, which is a per-net object, soserv->sv_nrthreadsis only counting the threads for its namespace.
OK. I missed the part where struct svc_serv is per-net, and each svc_serv has its own thread pool. Got it.
Quoted 12 lines
Each container that starts an nfsd will contribute "threads" number of nfsds to the pile. When you set "threads" to a different number, the total pile is grown or reduced accordingly. In fact, I'm not even sure we keep a systemwide total.
What _isn't_ done (unless I'm missing something) is any sort of limitation on the use of those threads. So you could have a container that starts one nfsd thread, but its clients can still have many more RPCs in flight, up to the total number of nfsd's running on the host. Limiting that might be possible, but I'm not yet convinced that it's a good idea.
Quoted 3 lines
It might be interesting to gather some metrics though. How often do the number of RPCs in flight exceed the number of threads that we started in a namespace? Might also be good to gather a high-water mark too?
I had a bunch of trace points queued up for this purpose, and all that got thrown out by Neil's recent work fixing up our pool thread scheduler.
I'm not sure that queued requests are all that meaningful or even quantifiable. A more meaningful metric is round trip time measured on clients.
Quoted 16 lines
Quoted 8 lines
Quoted 4 lines
By making this per-namespace as Josef suggests it should be fairly simple to tell whether my clients are regularly overrunning the threads I started. With this info as global, I have no idea what netns the RPCs being counted are against. I can't do anything with that info.
That's true. I'm just not sure we need to do anything significant about it as part of this patch series.
Quoted 7 lines
I'm not at all advocating leaving it like this. I agree it needs updating.
For now, just fill in that thread count stat with the global thread count, and later when we have a better concept for what it means to "adjust the nfsd thread count from inside a container" we can come back and make it make sense.
Quoted 6 lines
It would be good to step back and decide how we think this should work. What would this look like if we were designing it today? Then we can decide how to get there.
Personally, I think keeping the view inside the container as isolated as possible is the right thing to do.
I agree. I would like to get Josef's patches in quickly too.
Should that metric report svc_serv::sv_nrthreads instead of the global thread count?
--
Chuck Lever
Jeff Layton
maintainerRe: [PATCH v2 10/13] nfsd: move th_cnt into nfsd_net
Quoted 128 lines · Chuck Lever III
Quoted 104 lines · Jeff Layton
Quoted 90 lines · Chuck Lever III
Quoted 45 lines · Jeff Layton
Quoted 25 lines · Chuck Lever III
Quoted 14 lines · Jeff Layton
Quoted 9 lines · Josef Bacik
Quoted 8 lines · Chuck Lever
Quoted 2 lines · Josef Bacik
This is the last global stat, move it into
nfsd_netand adjust all the users to use that variant instead of the global one.Hm. I thought nfsd threads were a global resource -- they service all network namespaces. So, shouldn't the same thread count be surfaced to all containers? Won't they all see all of the nfsd processes?
Quoted 7 lines
Each container is going to start
/proc/fs/nfsd/threadsnumber of threads regardless. I hadn't actually grokked that they just get tossed onto the pile of threads that service requests.Is is possible for one container to start a small number of threads but have its client load be such that it spills over and ends up stealing threads from other containers?
I haven't seen any code that manages resources based on namespace, except in
filecache.cto restrict writeback per namespace.Quoted 8 lines
My impression is that any nfsd thread can serve any namespace. I'm not sure it is currently meaningful for a particular net namespace to "create" more threads.
If someone would like that level of control, we could implement a cgroup mechanism and have one or more separate
svc_poolsper net namespace, maybe? </hand wave>AFAICT, the total number of threads on the system will be the sum of the threads started in each of the containers. They do just go into a big pile, and whichever one wakes up will service the request, so the threads aren't associated with the netns, per-se. The svc_rqst's however _are_ per-netns. So, I don't see anything that ensures that a container doesn't exceed the number of threads it started on its own behalf.
Quoted 2 lines
<hand wave>
I'm not sure we'd need to tie this in to cgroups.Not a need, but cgroups are typically the way that such resource constraints are managed, that's all.
Quoted 38 lines
Now that Josef is
moving some of these key structures to be per-net, it should be fairly simple to havenfsd()just look at theth_cntand the thread count in the current namespace, and just enqueue the RPC rather than doing it? </hand wave>OTOH, maybe I'm overly concerned here.
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
I don't think we want the network namespaces seeing how many threads exist in the entire system right?
If someone in a non-init net namespace does a "pgrep -c nfsd" don't they see the total nfsd thread count for the host?
Yes, they're kernel threads and they aren't associated with a particular pid namespace.
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines
Additionally it appears that we can have multiple threads per network namespace, so it's not like this will just show 1 for each individual nn, it'll show however many threads have been configured for that nfsd in that network namespace.
Quoted 6 lines
I've never tried this, so I'm speculating. But it seems like for now, because all nfsd threads can serve all namespaces, they should all see the global thread count stat.
Then later we can refine it.
I don't think that info is particularly useful though, and it certainly breaks expectations wrt container isolation.
Look at it this way:
Quoted 9 lines
Quoted 3 lines
Suppose I have access to a container and I spin up nfsd with a particular number of threads. I now want to know "did I spin up enough threads?"
It makes sense to me for a container to ask for one or more NFSD listeners. But I'm not yet clear on what it means for a container to try to alter the NFSD thread count, which is currently global.
Quoted 3 lines
write_threadssets the number of nfsd threads for thesvc_serv, which is a per-net object, soserv->sv_nrthreadsis only counting the threads for its namespace.OK. I missed the part where struct
svc_servis per-net, and eachsvc_servhas its own thread pool. Got it.Quoted 12 lines
Each container that starts an nfsd will contribute "threads" number of nfsds to the pile. When you set "threads" to a different number, the total pile is grown or reduced accordingly. In fact, I'm not even sure we keep a systemwide total.
What _isn't_ done (unless I'm missing something) is any sort of limitation on the use of those threads. So you could have a container that starts one nfsd thread, but its clients can still have many more RPCs in flight, up to the total number of nfsd's running on the host. Limiting that might be possible, but I'm not yet convinced that it's a good idea.
Quoted 3 lines
It might be interesting to gather some metrics though. How often do the number of RPCs in flight exceed the number of threads that we started in a namespace? Might also be good to gather a high-water mark too?
Quoted 8 lines
I had a bunch of trace points queued up for this purpose, and all that got thrown out by Neil's recent work fixing up our pool thread scheduler.
I'm not sure that queued requests are all that meaningful or even quantifiable. A more meaningful metric is round trip time measured on clients.
I wouldn't try to queue anything just yet, but knowing when a particular container is exceeding the number of threads it started seems like a helpful metric. That could be done in a later patch though. I wouldn't block this work on that.
Quoted 21 lines
Quoted 21 lines
Quoted 9 lines
Quoted 5 lines
By making this per-namespace as Josef suggests it should be fairly simple to tell whether my clients are regularly overrunning the threads I started. With this info as global, I have no idea what netns the RPCs being counted are against. I can't do anything with that info.
That's true. I'm just not sure we need to do anything significant about it as part of this patch series.
Quoted 7 lines
I'm not at all advocating leaving it like this. I agree it needs updating.
For now, just fill in that thread count stat with the global thread count, and later when we have a better concept for what it means to "adjust the nfsd thread count from inside a container" we can come back and make it make sense.
It would be good to step back and decide how we think this should work. What would this look like if we were designing it today? Then we can decide how to get there.
Quoted 8 lines
Quoted 2 lines
Personally, I think keeping the view inside the container as isolated as possible is the right thing to do.
I agree. I would like to get Josef's patches in quickly too.
Should that metric report
svc_serv::sv_nrthreadsinstead of the global thread count?
No. The sv_nrthreads is just the same value that /proc/fs/nfsd/threads shows. The _metric_ should show the th_cnt for that particular namespace.
It would also be nice to know what the high-water mark for th_cnt is. IOW, if my containers clients exceeded the number of threads I've started, then it would be helpful to know by how much.
Signature · 1 line
Jeff Layton <[email protected]>