[SOLVED] Backup VM on NFS.

[divelgrom]

Hello. I have the following situation:
There are 2 Proxmox node in different countries. Between them built ipip tunnel.

Node A - Role: Active - NFS client - pve-manager/4.4-13/7ea56165 (running kernel: 4.4.49-1-pve)
Node B - Role: Backup\Standby - NFS server - pve-manager/4.4-13/7ea56165 (running kernel: 4.4.62-1-pve)

I make backup on node A in nfs. mode stop - working, mode snapshot and suspend - work endlessly.

During infinite execution, strace gives:

select(8, [5], NULL, NULL, {0, 254717}) = 0 (Timeout)
open("/proc/95946/stat", O_RDONLY) = 8
ioctl(8, SNDCTL_TMR_TIMEBASE or SNDRV_TIMER_IOCTL_NEXT_DEVICE or TCGETS, 0x7ffd24824fe0) = -1 ENOTTY (Inappropriate ioctl for device)
lseek(8, 0, SEEK_CUR) = 0
fstat(8, {st_mode=S_IFREG|0444, st_size=0, ...}) = 0
fcntl(8, F_SETFD, FD_CLOEXEC) = 0
read(8, "95946 (task UPIDve-0) S 95945 "..., 8192) = 328
close(8)

But for the vzdump process there is no 8 fd

# ps aux | grep vzdump
root 95945 1.6 0.0 262964 78508 pts/0 S+ 12:34 0:00 /usr/bin/perl -T /usr/bin/vzdump 166 --node pve-01 --mode snapshot --compress gzip --storage hetzner_backup_0 --remove 0 --tmpdir /tmp/

# ll /proc/95945/fd/
total 0
lrwx------ 1 root root 64 Jun 8 12:35 0 -> /dev/pts/0
lrwx------ 1 root root 64 Jun 8 12:35 1 -> /dev/pts/0
lr-x------ 1 root root 64 Jun 8 12:35 10 -> /usr/share/perl5/Net/LDAP/Constant.pm
lrwx------ 1 root root 64 Jun 8 12:35 2 -> /dev/pts/0
lr-x------ 1 root root 64 Jun 8 12:35 3 -> anon_inode:inotify
lrwx------ 1 root root 64 Jun 8 12:35 4 -> socket:[233235913]
lr-x------ 1 root root 64 Jun 8 12:35 5 -> pipe:[233235914]
l-wx------ 1 root root 64 Jun 8 12:35 6 -> /var/log/pve/tasks/0/UPIDve-01:000176CA:1AB946BE:59391A40:vzdump::root@pam:
lrwx------ 1 root root 64 Jun 8 12:35 7 -> socket:[233235921]

What could be the problem?

 

[manu]

are you able to write on the NFS share as backup hangs ? try to write small file and hopefully it does not hang

 

[manu]

btw it would be probably better to run the backup locally and use the hook functionnality of vzdump ( see the vzdump(1) man page)
to rsync the data to your second server NFS does not run very well over high latency links.

 

[divelgrom]

btw it would be probably better to run the backup locally and use the hook functionnality of vzdump ( see the vzdump(1) man page)
to rsync the data to your second server NFS does not run very well over high latency links.

Okey. I'll try, and the result will inform

 

[divelgrom]

As it turned out the problem is observed only with one virtual machine. Others can be backuped normally