Simplify rpcsec gss dependencies in unit files
Subsystems (2)
cover Simplify rpcsec gss dependencies in unit files
*** BLURB HERE ***
Concentrate dependencies for rpcsec GSS stuff in one place.
Simo Sorce (1):
Move the wants only to the auth unit.
systemd/auth-rpcgss-module.service | 3 ++- systemd/nfs-client.target | 4 ++-- systemd/nfs-server.service | 1 - 3 files changed, 4 insertions(+), 4 deletions(-)
Signature · 1 line
2.1.0
PATCH 1/1 Move the wants only to the auth unit.
This way either gssproxy or rpc.svcgssd are started only if the auth module is requested, and it finds a keytab.
If the wants are in the main nfs-client or nfs-server unit files then the two deamons are started unconditionally and would require conditions which we can test once and for all in a single unit file instead.
--- systemd/auth-rpcgss-module.service | 3 ++- systemd/nfs-client.target | 4 ++-- systemd/nfs-server.service | 1 - 3 files changed, 4 insertions(+), 4 deletions(-)
systemd/auth-rpcgss-module.service +2 -1
@@ -6,7 +6,8 @@ # unit will fail. But that's OK.) [Unit] Description=Kernel Module supporting RPCSEC_GSS -Before=gssproxy.service rpc-svcgssd.service +Before=gssproxy.service rpc-svcgssd.service rpc-gssd.service +Wants=gssproxy.service rpc-svcgssd.service rpc-gssd.service ConditionPathExists=/etc/krb5.keytab [Service]
systemd/nfs-client.target +2 -2
@@ -5,9 +5,9 @@ Wants=remote-fs-pre.target # Note: we don't "Wants=rpc-statd.service" as "mount.nfs" will arrange to # start that on demand if needed. -Wants=rpc-gssd.service rpc-svcgssd.service auth-rpcgss-module.service +Wants=auth-rpcgss-module.service Wants=nfs-blkmap.service rpc-statd-notify.service -Before=rpc-gssd.service rpc-svcgssd.service nfs-blkmap.service +Before=nfs-blkmap.service [Install] WantedBy=multi-user.target
systemd/nfs-server.service +0 -1
@@ -3,7 +3,6 @@ Description=NFS server and services Requires= network.target proc-fs-nfsd.mount rpcbind.target Requires= nfs-mountd.service Wants=rpc-statd.service nfs-idmapd.service auth-rpcgss-module.service -Wants=rpc-gssd.service gssproxy.service rpc-svcgssd.service Wants=rpc-statd-notify.service After= network.target proc-fs-nfsd.mount rpcbind.target nfs-mountd.service
Signature · 1 line
2.1.0
Re: [PATCH 1/1] Move the wants only to the auth unit.
Quoted 32 lines · Simo Sorce
This way either gssproxy or rpc.svcgssd are started only if the auth module is requested, and it finds a keytab.
If the wants are in the main nfs-client or nfs-server unit files then the two deamons are started unconditionally and would require conditions which we can test once and for all in a single unit file instead.Signed-off-by Simo Sorce <[email protected]>--- systemd/auth-rpcgss-module.service | 3 ++- systemd/nfs-client.target | 4 ++-- systemd/nfs-server.service | 1 - 3 files changed, 4 insertions(+), 4 deletions(-)systemd/auth-rpcgss-module.service +2
-1@@ -6,7 +6,8 @@ # unit will fail. But that's OK.) [Unit] Description=Kernel Module supporting RPCSEC_GSS -Before=gssproxy.service rpc-svcgssd.service +Before=gssproxy.service rpc-svcgssd.service rpc-gssd.service +Wants=gssproxy.service rpc-svcgssd.service rpc-gssd.service ConditionPathExists=/etc/krb5.keytab [Service]systemd/nfs-client.target +0
-0
systemd/nfs-client.target Wants=remote-fs-pre.target · line 10
Quoted 8 lines
systemd/nfs-client.target +2
-2@@ -5,9 +5,9 @@ Wants=remote-fs-pre.target # Note: we don't "Wants=rpc-statd.service" as "mount.nfs" will arrange to # start that on demand if needed. -Wants=rpc-gssd.service rpc-svcgssd.service auth-rpcgss-module.service +Wants=auth-rpcgss-module.service Wants=nfs-blkmap.service rpc-statd-notify.service -Before=rpc-gssd.service rpc-svcgssd.service nfs-blkmap.service +Before=nfs-blkmap.service
In the upstream version (commit 12a95edae) this is actually
After=rpc-gssd.service rpc-svcgssd.service nfs-blkmap.service
and I think we should leave it because sync things up nicely when both the server and client plus this patch is only about moving Wants around... not Afters...
steved.
Quoted 16 lines
systemd/nfs-client.target +0
-0[Install] WantedBy=multi-user.targetsystemd/nfs-server.service +0
-1@@ -3,7 +3,6 @@ Description=NFS server and services Requires= network.target proc-fs-nfsd.mount rpcbind.target Requires= nfs-mountd.service Wants=rpc-statd.service nfs-idmapd.service auth-rpcgss-module.service -Wants=rpc-gssd.service gssproxy.service rpc-svcgssd.service Wants=rpc-statd-notify.service After= network.target proc-fs-nfsd.mount rpcbind.target nfs-mountd.service
Re: [PATCH 1/1] Move the wants only to the auth unit.
Quoted 47 lines · On Tue, 30 Sep 2014 10:45:31 -0400 Steve Dickson <[email protected]> wrote:
Quoted 39 lines · Simo Sorce
This way either gssproxy or rpc.svcgssd are started only if the auth module is requested, and it finds a keytab.
If the wants are in the main nfs-client or nfs-server unit files then the two deamons are started unconditionally and would require conditions which we can test once and for all in a single unit file instead.Signed-off-by Simo Sorce <[email protected]>--- systemd/auth-rpcgss-module.service | 3 ++- systemd/nfs-client.target | 4 ++-- systemd/nfs-server.service | 1 - 3 files changed, 4 insertions(+), 4 deletions(-)(quoted diff) +0
-0b/systemd/auth-rpcgss-module.service index
3fc2f4ac924f7e9d6e24969bb9a21d88a5c144fc..0355e13e009528632e97373332db9fa3acdfd1a9100644 --- a/systemd/auth-rpcgss-module.service +++ b/systemd/auth-rpcgss-module.service @@ -6,7 +6,8 @@# unit will fail. But that's OK.) [Unit] Description=Kernel Module supporting RPCSEC_GSSsystemd/auth-rpcgss-module.service +2
-1-Before=gssproxy.service rpc-svcgssd.service +Before=gssproxy.service rpc-svcgssd.service rpc-gssd.service +Wants=gssproxy.service rpc-svcgssd.service rpc-gssd.service ConditionPathExists=/etc/krb5.keytab [Service]systemd/nfs-client.target +0
-0index
474f5e9ad74b70d265a7b2099556c2b9220bcdd0..f3d884a241543e41fdd91bcb0a055654b4b2c275100644 --- a/systemd/nfs-client.target +++ b/systemd/nfs-client.target @@ -5,9 +5,9 @@ Wants=remote-fs-pre.target
# Note: we don't "Wants=rpc-statd.service" as "mount.nfs" will
arrange to # start that on demand if needed.
-Wants=rpc-gssd.service rpc-svcgssd.service auth-rpcgss-module.service +Wants=
Wants=nfs-blkmap.service rpc-statd-notify.servicesystemd/nfs-client.target Wants=remote-fs-pre.target · line 10
Quoted 2 lines
systemd/nfs-client.target +1
-1-Before=rpc-gssd.service rpc-svcgssd.service nfs-blkmap.service +Before=nfs-blkmap.serviceIn the upstream version (commit
12a95edae) this is actually
After=rpc-gssd.service rpc-svcgssd.service nfs-blkmap.service
Quoted 3 lines
and I think we should leave it because sync things up nicely when both the server and client plus this patch is only about moving Wants around... not Afters...
I will send new patch with corrected After.
Simo.
Quoted 19 lines
steved.
Quoted 17 lines
[Install]
WantedBy=multi-user.targetsystemd/nfs-server.service +0
-0index
1048c5cbbf68328a8ac8c88b67e477061cf487c7..a5039062ad95ca0c73c8a6c33d053a845d596185100644 --- a/systemd/nfs-server.service +++ b/systemd/nfs-server.service @@ -3,7 +3,6 @@ Description=NFS server and services Requires= network.target proc-fs-nfsd.mount rpcbind.target Requires= nfs-mountd.service
Wants=rpc-statd.service nfs-idmapd.service
auth-rpcgss-module.service -Wants=rpc-gssd.service gssproxy.service rpc-svcgssd.service Wants=rpc-statd-notify.serviceAfter= network.target proc-fs-nfsd.mount rpcbind.target
nfs-mountd.service
Signature · 1 line
Simo Sorce * Red Hat, Inc * New York
Re: [PATCH 1/1] Move the wants only to the auth unit.
In reply to Simo SorceQuoted 18 lines · Simo Sorce
This way either gssproxy or rpc.svcgssd are started only if the auth module is requested, and it finds a keytab.
If the wants are in the main nfs-client or nfs-server unit files then the two deamons are started unconditionally and would require conditions which we can test once and for all in a single unit file instead.Signed-off-by Simo Sorce <[email protected]>--- systemd/auth-rpcgss-module.service | 3 ++- systemd/nfs-client.target | 4 ++-- systemd/nfs-server.service | 1 - 3 files changed, 4 insertions(+), 4 deletions(-)systemd/auth-rpcgss-module.service +0
-0
systemd/auth-rpcgss-module.service line 9
Quoted 5 lines
systemd/auth-rpcgss-module.service +1
-1@@ -6,7 +6,8 @@ # unit will fail. But that's OK.) [Unit] Description=Kernel Module supporting RPCSEC_GSS -Before=gssproxy.service rpc-svcgssd.service +Before=gssproxy.service rpc-svcgssd.service rpc-gssd.service
By moving these into this unit,it destroys client/server sync starts commit 12a95eda talks about...
Maybe we could put an After=nfs-server.service in nfs-client.target to bring back that synchronization... because in the end we really really want the server to start first especially when gssproxy is involved and both units are enabled.
Also, this will cause gssproxy to be started on every boot regardless whether Kerberos is installed and configured (which not the case with rpc.svcgssd)...
I can hear the complaints already... Why is NFS starting up this daemon that will never have anything to do, in the case when Kerberos is not installed/configure which could be the majority of the cases...
I would really really, really like to avoid this.
steved.
Quoted 33 lines
systemd/auth-rpcgss-module.service +1
-0+Wants=gssproxy.service rpc-svcgssd.service rpc-gssd.service ConditionPathExists=/etc/krb5.keytab [Service]systemd/nfs-client.target +2
-2@@ -5,9 +5,9 @@ Wants=remote-fs-pre.target # Note: we don't "Wants=rpc-statd.service" as "mount.nfs" will arrange to # start that on demand if needed. -Wants=rpc-gssd.service rpc-svcgssd.service auth-rpcgss-module.service +Wants=auth-rpcgss-module.service Wants=nfs-blkmap.service rpc-statd-notify.service -Before=rpc-gssd.service rpc-svcgssd.service nfs-blkmap.service +Before=nfs-blkmap.service [Install] WantedBy=multi-user.targetsystemd/nfs-server.service +0
-1@@ -3,7 +3,6 @@ Description=NFS server and services Requires= network.target proc-fs-nfsd.mount rpcbind.target Requires= nfs-mountd.service Wants=rpc-statd.service nfs-idmapd.service auth-rpcgss-module.service -Wants=rpc-gssd.service gssproxy.service rpc-svcgssd.service Wants=rpc-statd-notify.service After= network.target proc-fs-nfsd.mount rpcbind.target nfs-mountd.service
Re: [PATCH 1/1] Move the wants only to the auth unit.
Quoted 15 lines · On Tue, 30 Sep 2014 11:05:14 -0400 Steve Dickson <[email protected]> wrote:
Quoted 14 lines · Simo Sorce
This way either gssproxy or rpc.svcgssd are started only if the auth module is requested, and it finds a keytab.
If the wants are in the main nfs-client or nfs-server unit files then the two deamons are started unconditionally and would require conditions which we can test once and for all in a single unit file instead.Signed-off-by Simo Sorce <[email protected]>--- systemd/auth-rpcgss-module.service | 3 ++- systemd/nfs-client.target | 4 ++-- systemd/nfs-server.service | 1 - 3 files changed, 4 insertions(+), 4 deletions(-)
Quoted 12 lines
systemd/auth-rpcgss-module.service line 9
Quoted 10 lines
(quoted diff) +0
-0b/systemd/auth-rpcgss-module.service index
3fc2f4ac924f7e9d6e24969bb9a21d88a5c144fc..0355e13e009528632e97373332db9fa3acdfd1a9100644 --- a/systemd/auth-rpcgss-module.service +++ b/systemd/auth-rpcgss-module.service @@ -6,7 +6,8 @@# unit will fail. But that's OK.) [Unit] Description=Kernel Module supporting RPCSEC_GSSsystemd/auth-rpcgss-module.service +1
-1-Before=gssproxy.service rpc-svcgssd.service +Before=gssproxy.service rpc-svcgssd.service rpc-gssd.serviceBy moving these into this unit,it destroys client/server sync starts commit
12a95edatalks about...
No it does not, this before is critical, the kernel module must be loaded before the gss daemons are started.
Quoted 1 line
Maybe we could put an After=nfs-server.service in nfs-client.target
Why would you load the auth modules *after* nfs client and servers are started ?
I think this could cause race conditions at mount on boot if someone wants to mount a filesystem with sec=krb5
Quoted 3 lines
to bring back that synchronization... because in the end we really really want the server to start first especially when gssproxy is involved and both units are enabled.
uh ?
no you really want to start the auth damoens first, if the server starts first then a mount request from a client could race with gssproxy starting up and poking the proc file to enable use of gssproxy resulting in the auth module to permanently initialize to use the old interface.
Quoted 3 lines
Also, this will cause gssproxy to be started on every boot regardless whether Kerberos is installed and configured (which not the case with rpc.svcgssd)...
This is a separate issue, I am trying to get the ordering right here, then we can debug why gssproxy is started, I suspect Wants always kicks in regrardless of ConditionPathExists.
Quoted 6 lines
I can hear the complaints already... Why is NFS starting up this daemon that will never have anything to do, in the case when Kerberos is not installed/configure which could be the majority of the cases...
I would really really, really like to avoid this.
We can handle it in the next patch. Let's get the order right first.
Simo.
Signature · 1 line
Simo Sorce * Red Hat, Inc * New York
Re: [PATCH 1/1] Move the wants only to the auth unit.
Quoted 31 lines · Simo Sorce
Quoted 15 lines · On Tue, 30 Sep 2014 11:05:14 -0400 Steve Dickson <[email protected]> wrote:
Quoted 14 lines · Simo Sorce
This way either gssproxy or rpc.svcgssd are started only if the auth module is requested, and it finds a keytab.
If the wants are in the main nfs-client or nfs-server unit files then the two deamons are started unconditionally and would require conditions which we can test once and for all in a single unit file instead.Signed-off-by Simo Sorce <[email protected]>--- systemd/auth-rpcgss-module.service | 3 ++- systemd/nfs-client.target | 4 ++-- systemd/nfs-server.service | 1 - 3 files changed, 4 insertions(+), 4 deletions(-)Quoted 12 lines
systemd/auth-rpcgss-module.service line 9
Quoted 10 lines
(quoted diff) +0
-0b/systemd/auth-rpcgss-module.service index
3fc2f4ac924f7e9d6e24969bb9a21d88a5c144fc..0355e13e009528632e97373332db9fa3acdfd1a9100644 --- a/systemd/auth-rpcgss-module.service +++ b/systemd/auth-rpcgss-module.service @@ -6,7 +6,8 @@# unit will fail. But that's OK.) [Unit] Description=Kernel Module supporting RPCSEC_GSSsystemd/auth-rpcgss-module.service +1
-1-Before=gssproxy.service rpc-svcgssd.service +Before=gssproxy.service rpc-svcgssd.service rpc-gssd.serviceBy moving these into this unit,it destroys client/server sync starts commit
12a95edatalks about...
Quoted 2 lines
No it does not, this before is critical, the kernel module must be loaded before the gss daemons are started.
Understood.... but both the nfs-server.service and nfs-client.target units have a Wants=auth-rpcgss-module.service, is not clear unit will get started first... On one of my very fast booting machines this race caused ordering cycles in systemd... I think... at least when I ordered the server to start then the client the cycles went a way... but who knows...
Quoted 7 lines
Quoted 2 lines
Maybe we could put an After=nfs-server.service in nfs-client.target
Why would you load the auth modules *after* nfs client and servers are started ?
I think this could cause race conditions at mount on boot if someone wants to mount a filesystem with sec=krb5
Again it has nothing to do the loading of the module... You are correct the they have to be loaded before the gss daemons are started. Its all about the ordering of the server and client units.
Quoted 11 lines
Quoted 4 lines
to bring back that synchronization... because in the end we really really want the server to start first especially when gssproxy is involved and both units are enabled.
uh ?
no you really want to start the auth damoens first, if the server starts first then a mount request from a client could race with gssproxy starting up and poking the proc file to enable use of gssproxy resulting in the auth module to permanently initialize to use the old interface.
My bad on this one... I was thinking about when the nfs-client.target was not starting gssproxy... there was race with the server which caused both gssproxy and rpc.svcgssd to be started. Now that the client is starting gssproxy, that race no longer exists...
Quoted 9 lines
Quoted 4 lines
Also, this will cause gssproxy to be started on every boot regardless whether Kerberos is installed and configured (which not the case with rpc.svcgssd)...
This is a separate issue, I am trying to get the ordering right here, then we can debug why gssproxy is started, I suspect Wants always kicks in regrardless of ConditionPathExists.
Quoted 8 lines
Quoted 6 lines
I can hear the complaints already... Why is NFS starting up this daemon that will never have anything to do, in the case when Kerberos is not installed/configure which could be the majority of the cases...
I would really really, really like to avoid this.
We can handle it in the next patch. Let's get the order right first.
Fine...
steved.
Quoted 5 lines
Simo.
Re: [PATCH 1/1] Move the wants only to the auth unit.
Quoted 34 lines · On Tue, 30 Sep 2014 11:54:47 -0400 Steve Dickson <[email protected]> wrote:
Quoted 31 lines · Simo Sorce
Quoted 27 lines · On Tue, 30 Sep 2014 11:05:14 -0400 Steve Dickson <[email protected]> wrote:
Quoted 22 lines · Simo Sorce
This way either gssproxy or rpc.svcgssd are started only if the auth module is requested, and it finds a keytab.
If the wants are in the main nfs-client or nfs-server unit files then the two deamons are started unconditionally and would require conditions which we can test once and for all in a single unit file instead.Signed-off-by Simo Sorce <[email protected]>--- systemd/auth-rpcgss-module.service | 3 ++- systemd/nfs-client.target | 4 ++-- systemd/nfs-server.service | 1 - 3 files changed, 4 insertions(+), 4 deletions(-)(quoted diff) +0
-0b/systemd/auth-rpcgss-module.service index
3fc2f4ac924f7e9d6e24969bb9a21d88a5c144fc..0355e13e009528632e97373332db9fa3acdfd1a9100644 --- a/systemd/auth-rpcgss-module.service +++ b/systemd/auth-rpcgss-module.service @@ -6,7 +6,8 @@# unit will fail. But that's OK.) [Unit] Description=Kernel Module supporting RPCSEC_GSSsystemd/auth-rpcgss-module.service line 9
Quoted 2 lines
systemd/auth-rpcgss-module.service +1
-1-Before=gssproxy.service rpc-svcgssd.service +Before=gssproxy.service rpc-svcgssd.service rpc-gssd.serviceBy moving these into this unit,it destroys client/server sync starts commit
12a95edatalks about...
Quoted 8 lines
Quoted 2 lines
No it does not, this before is critical, the kernel module must be loaded before the gss daemons are started.
Understood.... but both the nfs-server.service and nfs-client.target units have a Wants=auth-rpcgss-module.service, is not clear unit will get started first... On one of my very fast booting machines this race caused ordering cycles in systemd... I think... at least when I ordered the server to start then the client the cycles went a way... but who knows...
Both nfs-server and nfs-client unit files have After: <gss unit files> so they are always after them (I fixed a missing gssproxy.service one in the last patch)
And auth-rpcgss-module.service has Before: <gss unit files>
So the ordering is fixed as per my last patch commit message.
Quoted 10 lines
Quoted 7 lines
Quoted 2 lines
Maybe we could put an After=nfs-server.service in nfs-client.target
Why would you load the auth modules *after* nfs client and servers are started ?
I think this could cause race conditions at mount on boot if someone wants to mount a filesystem with sec=krb5Again it has nothing to do the loading of the module... You are correct the they have to be loaded before the gss daemons are started. Its all about the ordering of the server and client units.
So what is the ordering you are concerned about ?
The way I understand it is
1. load module
----
2. start rpc.gssd AND (rpc-svcgssd.service OR gssproxy.service) order between these seem not important so they can start in parallel
----
3. nfs client AND/OR nfs server
I do not know if there needs to be any ordering between the 2 above, I operated on the assumption they can start in parallel
I left any ancillary daemon (statd/imapd/etc...) as it were, but if you can add the ordering here I can double check those are starting in the right order too.
Quoted 11 lines
Quoted 11 lines
Quoted 4 lines
to bring back that synchronization... because in the end we really really want the server to start first especially when gssproxy is involved and both units are enabled.
uh ?
no you really want to start the auth damoens first, if the server starts first then a mount request from a client could race with gssproxy starting up and poking the proc file to enable use of gssproxy resulting in the auth module to permanently initialize to use the old interface.
Quoted 4 lines
My bad on this one... I was thinking about when the nfs-client.target was not starting gssproxy... there was race with the server which caused both gssproxy and rpc.svcgssd to be started. Now that the client is starting gssproxy, that race no longer exists...
ok.
please let me know if I wrongly understood the ordering requirements we have.
Simo.
Signature · 1 line
Simo Sorce * Red Hat, Inc * New York