[ClusterLabs] Q: Is there any plan for pcs to support corosync-notifyd?

[井上和徳]

On Tue, Mar 16, 2021 at 11:24 PM Ken Gaillot <kgaillot at redhat.com> wrote:
>
> On Tue, 2021-03-16 at 13:24 +0100, damiano giuliani wrote:
> > Could be an idea to  let your cluster start it on all your nodes
> >
> > Br
>
> That does sound like a good idea. That way the cluster can monitor it
> (with the equivalent of systemctl status) and restart it if needed, and
> constraints can be used if some other resource depends on it. Something
> like:
>
> pcs resource create notifyd systemd:corosync-notifyd clone

I appreciate your proposal.

With this method, the trap at corosync startup is hardly sent. I will ask the
user for their opinion.
# pcs cluster start --all
rhel83-1: Starting Cluster...
rhel83-2: Starting Cluster...
rhel83-3: Starting Cluster...

Mar 18 12:48:47 cent83 snmptrapd[32754]: (snip) rhel83-2 (snip)
STRING: "quorate"
Mar 18 12:48:47 cent83 snmptrapd[32754]: (snip) rhel83-3 (snip)
STRING: "quorate"
Mar 18 12:48:47 cent83 snmptrapd[32754]: (snip) rhel83-1 (snip)
STRING: "quorate"

For your information, when corosync and notifyd are started at the same time,
the following trap is sent.
# for i in rhel83-{1..3}; do ssh -f $i 'systemctl start corosync-notifyd'; done

Mar 18 16:24:07 cent83 snmptrapd[33312]: (snip) rhel83-2 (snip)
STRING: "operational"
Mar 18 16:24:07 cent83 snmptrapd[33312]: (snip) rhel83-2 (snip)
STRING: "operational"
Mar 18 16:24:07 cent83 snmptrapd[33312]: (snip) rhel83-3 (snip)
STRING: "operational"
Mar 18 16:24:07 cent83 snmptrapd[33312]: (snip) rhel83-3 (snip)
STRING: "operational"
Mar 18 16:24:07 cent83 snmptrapd[33312]: (snip) rhel83-3 (snip)
STRING: "operational"
Mar 18 16:24:07 cent83 snmptrapd[33312]: (snip) rhel83-3 (snip)
STRING: "operational"
Mar 18 16:24:08 cent83 snmptrapd[33312]: (snip) rhel83-1 (snip)
STRING: "operational"
Mar 18 16:24:08 cent83 snmptrapd[33312]: (snip) rhel83-1 (snip)
STRING: "operational"
Mar 18 16:24:08 cent83 snmptrapd[33312]: (snip) rhel83-1 (snip)
STRING: "operational"
Mar 18 16:24:08 cent83 snmptrapd[33312]: (snip) rhel83-1 (snip)
STRING: "operational"
Mar 18 16:24:08 cent83 snmptrapd[33312]: (snip) rhel83-2 (snip)
STRING: "operational"
Mar 18 16:24:08 cent83 snmptrapd[33312]: (snip) rhel83-2 (snip)
STRING: "operational"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-1 (snip) STRING: "joined"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-1 (snip)
STRING: "quorate"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-1 (snip) STRING: "joined"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-3 (snip) STRING: "joined"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-3 (snip)
STRING: "quorate"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-3 (snip) STRING: "joined"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-2 (snip) STRING: "joined"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-2 (snip)
STRING: "quorate"
Mar 18 16:24:11 cent83 snmptrapd[33312]: (snip) rhel83-2 (snip) STRING: "joined"

> > On Tue, 16 Mar 2021, 09:58 井上和徳, <kazunori_inoue at newson.co.jp> wrote:
> > > Hi!
> > >
> > > Cluster (corosync and pacemaker) can be started with pcs,
> > > but corosync-notifyd needs to be started separately with systemctl,
> > > which is not easy to use.
> > >
> > > # pcs cluster start --all
> > > rhel83-1: Starting Cluster...
> > > rhel83-2: Starting Cluster...
> > > rhel83-3: Starting Cluster...
> > > # ssh rhel83-1 systemctl start corosync-notifyd
> > > # ssh rhel83-2 systemctl start corosync-notifyd
> > > # ssh rhel83-3 systemctl start corosync-notifyd
> > >
> > > Is there any plan for pcs to support corosync-notifyd?
> > >
> > > Regards,
> > > Kazunori INOUE
> --
> Ken Gaillot <kgaillot at redhat.com>