Full Sync after First node restart

[simlac]

Hi all,

I have a three nodes cluster that I'm implementing : two data nodes and a witness.
Everything is running smoothly, except when I try to reboot the "First" node.

When it comes back online, all the devices that are synchronised start a full sync.
When it's the "Second" node that I restart, it only starts a delta sync, as well as when I'm in Heartbeat replication without witness.
I do not put them in maintenance mode, as I want to simulate a node failure.

Is there a way to avoid that full sync ?

Thanks !

[yaroslav (staff)]

Greetings please collect logs from all 3 boxes and share them with me https://knowledgebase.starwindsoftware. ... collector/. Share them via Google Drive. Make sure to follow this reboot procedure https://knowledgebase.starwindsoftware. ... installed/

It can be that 1 is the only active server while 2 is not synchronized.

[simlac]

Thanks for your quick answer !
You'll find the files here https://drive.google.com/drive/folders/ ... sp=sharing

As for the shutdown procedure, everything is working great when I follow it, but I would like the experiment the case of a node crash.

Thanks again,

[yaroslav (staff)]

I believe that I found the problem. Please wait for all the HA devices to synchronize. I found that the last time service was restarted while other side was not synchornized.
Also, I noticed that at 14:20 on 15/01 service was stopped on 02 while at 14:21 it was stopped on 2 other nodes. Full sync here is expected as StarWind Service is stopped on all 3 nodes.
Please make sure to test an unexpected shutdown of a server as described here https://knowledgebase.starwindsoftware. ... installed/ (instead of shut down gracefully there should be an unexpected shutdown in your case).

If you aim to test the unexpected shutdown of all 3 servers though, full synchronization is expected in that case. You can avoid that process though by putting StarWind HA devices into maintenance mode as discussed in the article I shared.

[simlac]

Thanks once again for your work !

I think I had the issue even with full synced devices, and a only the first node shut down.
I will try again with this configuration, and the shutdown documentation you provided.

I'll keep you in the loop about it, and provide the logs if necessary.

Cheers,

[yaroslav (staff)]

Yes, please keep me posted.

[simlac]

Hi again !
I recreated the issue, with a Device "testRebootW" fully synced (the other devices are left in maintenance mode)
Both the GUI and the CLI shows the device as synced.
At around 10:10, I stopped the server hosting the device noted as "First", and on boot, a full synchro started.

You'll find the generated logs here
https://drive.google.com/drive/folders/ ... sp=sharing

Thanks again,
Cheers

[yaroslav (staff)]

Greetings,

Could you log a case with us? Fill in this form https://www.starwindsoftware.com/support-form.
Use 444545 and this forum thread as your references.

[simlac]

Done !
I defeated the evil blue reset button.
Once again, thanks for your work.

Cheers !

[yaroslav (staff)]