2.2 my first snapshot test

[Jae Lee]

hi so i just upgraded from 2.1 to 2.2 w/o a problem and i created a new KVM w/ qcow2, installed cpanel, and then went to proxmox to try out the new snapshot.
now it's been over 5 mins and still says: Task: VM 100 - Snapshot w/ the progress bar looping over and over.
and I see that the VM is not responding.
is this really how it's supposed to work?

 

[Jae Lee]

it's almost over 15 minutes now, seems it just froze. with no abort button, what can i do?
on top of that, i'm looking at /var/lib/vz/images/100, and i see nothing there!
did snapshot feature somehow delete it?

 

[Jae Lee]

now after waiting a bit longer it finally timed out with this message:

Formatting '/storage/hit2tb/images/100/vm-100-state-first.raw', fmt=raw size=17553162240
TASK ERROR: VM 100 qmp command 'snapshot-drive' failed - got timeout

any ideas?

and btw, VM is responsive again.

 

[dietmar]

Seems to be a problem with the storage on /storage/hit2tb - allocating a 16GB file should be fast, but takes forever?

BTW, you can stop any task - just open the task log view and press 'stop'

 

[Jae Lee]

hm, i just did a smartctl test and everything was fine on that drive /storage/hit2tb (it is a mounted drive btw).
it's formatted as ext4.
what could the problem be? anyway to find out?

 

[dietmar]

what could the problem be? anyway to find out?

Is the behavior reproducible?

 

[dietmar]

OK, seem your VM uses 8GB RAM. If the VM is running, that means the we need save up to 16GB RAM state.
Seems you storage is simply to slow to save that withing 10 minutes (hardcoded timeout). What kind of storage is that (how fast)?

 

[dietmar]

Seems you storage is simply to slow to save that withing 10 minutes (hardcoded timeout).

Ignore me - that conclusion was wrong. Instead, please can you tell me the size of that qcow2 disk?

 

[Jae Lee]

it's a 72k rpm sata.

hit2tb:100/vm-100-disk-1.qcow2,cache=writethrough,size=320G

 

[Jae Lee]

i just tried snapshot again, but same result.
but to give you some more insights, after i upgraded from 2.1 to 2.2, i didn't reboot.
could that be the problem?

also, i don't see anyway to cancel this, even in tasks,

 

[Jae Lee]

so i just rebooted and this time snapshot worked. but during the time to snapshot, server was un-responsive for the whole 2-3 minutes.
is this how it's supposed to be?

 

[tom]

no.

 

[dietmar]

but during the time to snapshot, server was un-responsive for the whole 2-3 minutes.

What do you mean by 'unresponsive'?

 

[Jae Lee]

basically the VM was offline, all my sites on it were down, couldn't ssh/ping, nothing was working.

 

[Jae Lee]

could this be happening because the storage isn't on LVM? currently it's using just a ext4 mounted folder.

 

[dietmar]

could this be happening because the storage isn't on LVM? currently it's using just a ext4 mounted folder.

Currently, live snapshot only works with qcow2 files, and that does not work on LVM (needs a file system). But It seems that qcow2 is very slow on creating snapshots on large files - while it works perfectly on smaller files.

 

[tom]

how fast is your ext4 storage? run pveperf and post it here:

> pveperf /path-to-your-ext4-storage

 

[eSportler]

Hey guys, i've got the same Problem as Jae Lee.

The output of pveperf /home ist

CPU BOGOMIPS: 54401.04
REGEX/SECOND: 1712089
HD SIZE: 1740.20 GB (/dev/md3)
BUFFERED READS: 145.22 MB/sec
AVERAGE SEEK TIME: 14.30 ms
FSYNCS/SECOND: 51.08
DNS EXT: 55.28 ms

Any Idea how to fix that the vms are not reachable while taking a snapshot?

Thanks and best regards
Carsten