Corosync user error

[jordan_richardo]

Hello there,

I was happily following along in the wiki at https://pve.proxmox.com/wiki/Separate_Cluster_Network
to change our Corosync network over to another subnet, all was going well until I realized that I skipped over the part about
"config_version: increase it, very important, you can write any number which is higher then the actual one, but you need to increase it."

I am having issues now trying to access the file and edit it and receive errors that it is locked for editing, as I assume Chorum is not able to tell which file is the master. I have tried stopping the cluster service but the file is of course no longer in that location.

Can someone help point me in the right direction? My assumption is that I need to have my other nodes offline, edit the file on the primary somehow before turning the nodes back online?

...I dun goofed lol

 

[wolfgang]

Hi,

please send you corosync.conf to identify the error.

 

[t.lamprecht]

Can someone help point me in the right direction? My assumption is that I need to have my other nodes offline, edit the file on the primary somehow before turning the nodes back online?

What you normally would do is:
* get a copy of corosync.conf which is good (e.g., the one before you started changing it if you're unsure, else you can also use the one you wanted to setup with the separate networks).
* setup a config_version number which is higher than all previously edited ones.
* copy it to all cluster nodes, but not to /etc/pve/corosync.conf but to /etc/corosync/corosync.conf (this is the location where corosync loads it's initial config on startup, the /etc/pve one then gets synced over on changes)
* restart corosync and pve-cluster on all cluster nodes:

systemctl restart corosync pve-cluster

now the notes should be quorate again, as final step copy the now blessed/active corosync config also to the now writeable /etc/pve/

This assumes you already lost quorum, did not use HA or used HA but the watchdog fenced already all HA nodes and thus you do not have to lose much more anymore.
If this is not your situation then maybe try do explain your current state a bit better and do not just blindly follow above steps, as then you can still make it worse

 

[jordan_richardo]

Good morning!

Thank you t.lamprecht for the insight. I ended up doing something similar last night when I should have just sat on my hands like kindergarden class lol.

I had stopped corosync and pve-cluster on each node, updated the file at /etc/pve/corosync.conf (realizing that this is the wrong file now lol), incremented the version string and restarted each node one at a time.

Everything came back up and is working as expected now which is great, but I am wondering if I should be editing that /etc/corosync/corosync.conf as well at this point?

Thank you,
Jordan

 

[t.lamprecht]

Good on you that you could got it up again yourself!

Everything came back up and is working as expected now which is great, but I am wondering if I should be editing that /etc/corosync/corosync.conf as well at this point?

Normally, as the /etc/pve one got a higher config_version number than the /etc/corosync one pmxcfs (the binary behind pve-cluster) should have synced it over again automatically.

But actually cou can just compare them to see if they differ:

diff -u /etc/pve/corosync.conf /etc/corosync/corosync.conf

IT should return nothing an your're good, else check the content of both and copy the correct (newest and valid) one over the other.

 

[jordan_richardo]

Thank you again t.lanprecht!

I did the compare as you suggested and returned nothing

I'm one of those people that can't leave well enough alone though, so it definitely could have went the other direction as well haha.

Thank you very much for taking the time with me,
Jordan