Why is my Vzdump hanging?

[jimmyjoe]

I have about 15 VMs I'm vzdump-ing to a sata RAID array every night around 2am. This morning I see vzdump is stopped. It has been about 7 hours since the vzdump started. Why? Here is the process list of vzdump (running since 2am). My backup process is using "snapshot" mode so I think this must be the problem. I will change to "suspend" like I had to do on another server but hopefully this can get fixed because suspend mode makes my DNS services go away during backup:

# ps -ef | grep dump
root      844476  844475  0 01:59 ?        00:00:00 /bin/sh -c vzdump --quiet 1 --mode snapshot --all 1 --node piston --compress lzo --storage local
root      844477  844476  0 01:59 ?        00:00:07 /usr/bin/perl -w -T /usr/bin/vzdump --quiet 1 --mode snapshot --all 1 --node piston --compress lzo --storage local
root      844478  844477  0 01:59 ?        00:00:05 task UPID:piston:000CE2BE:0686A469:55092271:vzdump::root@pam:
root      849459  844478  0 02:45 ?        00:00:48 /usr/bin/vma create -v -c /var/lib/vz/dump/vzdump-qemu-111-2015_03_18-02_45_58.tmp/qemu-server.conf exec:lzop>/var/lib/vz/dump/vzdump-qemu-111-2015_03_18-02_45_58.vma.dat drive-ide0=/var/lib/ssd/images/111/base-111-disk-1.raw
root      849460  849459  0 02:45 ?        00:00:00 sh -c lzop>/var/lib/vz/dump/vzdump-qemu-111-2015_03_18-02_45_58.vma.dat
root      895116  783641  0 09:56 pts/0    00:00:00 grep dump

Here is the log for qemu-111 VM:

# cat /var/log/vzdump/qemu-111.log 
Mar 18 02:45:58 INFO: Starting Backup of VM 111 (qemu)
Mar 18 02:45:58 INFO: status = stopped
Mar 18 02:45:58 INFO: update VM 111: -lock backup
Mar 18 02:45:58 INFO: backup mode: stop
Mar 18 02:45:58 INFO: ionice priority: 7
Mar 18 02:45:58 INFO: creating archive '/var/lib/vz/dump/vzdump-qemu-111-2015_03_18-02_45_58.vma.lzo'
Mar 18 02:45:58 INFO: starting template backup
Mar 18 02:45:58 INFO: /usr/bin/vma create -v -c /var/lib/vz/dump/vzdump-qemu-111-2015_03_18-02_45_58.tmp/qemu-server.conf exec:lzop>/var/lib/vz/dump/vzdump-qemu-111-2015_03_18-02_45_58.vma.dat drive-ide0=/var/lib/ssd/images/111/base-111-disk-1.raw

 

[jimmyjoe]

I ended up disabling the Proxmox backup job in the web GUI. For one thing, It tries to restart my Template hosts when backing them up and then considers it a failure because it can't "restart" them (??). The other problem was the hanging vzdump processes. I would log into a host ever couple days to find several just sitting there doing nothing. The process would die for some reason and never perform any backups for days.

I gave up and just use a custom vzdump script launched from cron. It basically loops through the list of VMs returned fro `qm list` and does this. Hope it saves someone hassle:

vzdump $vm --dumpdir /var/lib/vz/dump/ --mailto root --compress lzo --mailnotification failure --ionice 3 --mode snapshot --size 1000 --maxfiles 3