I'm dealing with a frustrating issue after recovering from a cluster failure caused by a failed HDD on my CentOS 6.7 setup. After replacing the failed drive, I tried to reintroduce it to the mirror, but now the synchronization with DRBD is extremely slow. I have two identical servers set up, both using mirrored drives and configured to work with DRBD. However, while most of my other clusters are syncing without any issues, this particular clustering is lagging significantly. I've checked the settings, swapped out the Ethernet cable, and even rebooted, but nothing seems to speed things up. I'm getting really low sync speeds and there seems to be corruption on the secondary drive. I want to know what else I can check or adjust to improve the situation. Any suggestions?
1 Answer
Hey! It sounds like you've got a solid grasp of the situation already. You might want to try adjusting the `max-buffers` setting. Setting it to `20000` could help with the resync speed. Just make sure you're applying this on both servers because in DRBD 8.4 a lot of the speed negotiation happens on the Secondary node too.
I've set `max-buffers` to `20480` on both sides, but the sync speed is still really slow, like around 8 KB/s most of the time. What else do I need to check?