[SOLVED] pvecm status fails to parse valid corosync.conf (missing ':' after key '3')

[cluck]

# pvecm status
missing ':' after key '2'
Can't use an undefined value as a HASH reference at /usr/share/perl5/PVE/CLI/pvecm.pm line 486, <DATA> line 960.

This is while corosync is happily running, `corosync-cfgtool -n` shows running links etc..

Looks to me like pvecm is choking on the parameter auto_tie_breaker_node which accepts a multi-token argument value.

The snipped below is from the votequorum(5) man page:

quorum {
provider: corosync_votequorum
expected_votes: 8
auto_tie_breaker: 1
auto_tie_breaker_node: 1 3 5
}

 

[Alwin Antreich]

Compare /etc/pve/corosync.conf with /etc/corosync/corosyn.conf. The one in /etc/pve is the one to edit, as it will be distributed across the cluster, when it is quorate.

Other then that the whole nodelist is missing, aside from other parts.

 

[cluck]

My corosync is working properly. I haven't posted my corosync.conf because I think I see a parser bug that is independent of the config file in /usr/share/perl5/PVE/Corosync.pm (approx line 92).

The parser is just a tad too simplistic and fails to parse valid corosync.conf files when they contain multi-valued options. E.g. it fails to parse "auto_tie_breaker_node: 1 3 5"), but succeeds in parsing "auto_tie_breaker_node: 3". The configuration snipped posted previously is just the "part" of corosync.conf required to reproduce the bug, and to give some minimal context.

 

[cluck]

Now that I walked into it, I have reported it in Bugzilla. I previously ended up reporting it here because I sloppily googled for "how to report bug proxmox", which led to the getting help for Proxmox page, which refers to this forum.