Harddisk free space reclaiming and i/o errors

minblock commented 4 years ago

My harddrive pool didnt resize as expected when I deleted snapshots and vms, I stopped them all and the space reclaimed. An orphaned snapshot that I restored had a main drive space of 12GB and had only 4MB of space remaining. Its original VM was deleted. This might of been caused by the configuration being curropted or deleted with the original owner of the snapshot. I am also using a LVM storage for all my harddrives spanning over 3 now. The HD Pool space might be thin provisioning im not sure. It did run out of space mid use and crashed the entire xenserver causing a ripple effect. During the expanding the main vg process, (it was full) fdisk had an error about the drive partition being out of bounds and a write operation would fix this. So what I am asking is the lvm storage pool the cause of my I/O errors in all vms. This could be extremely exploitable as I wouldnt be able to offer Storage services to anyone. The message log is all gzips. What would be a log to inspect?

nagilum99 commented 4 years ago

Normally you shouldn't be able to overprovision (and possibly run out of space with) LVM. It should cancel the operation with an error! Thin provisioning with LVM has never been implemented with XS and AFAIK neither with XCP-ng, so that would be a bug.

stormi commented 4 years ago

Hi. Sorry for now answering sooner. I don't know what to answer with the available data. If you still need to investigate this issue, the options are:

read https://github.com/xcp-ng/xcp/wiki/Troubleshooting
ask our forum
ask our pro support

stormi commented 3 years ago

Closing as we don't use github issues as a support channel. Feel free to reopen if the logs reveal a bug.

xcp-ng / xcp

Harddisk free space reclaiming and i/o errors #278