Proxmox Backups - An option to not use HCP Snapshots

[marotori]

Latest kernel is 2.6.32-10-pve

I use this and a custome lsi megaraid driver for all mine.

 

[bazzi]

[EMAIL="root@timo"]root@timo[/EMAIL]:~# uname -r
2.6.32-10-pve
[EMAIL="root@timo"]root@timo[/EMAIL]:~# lsmod |grep sas
mptsas                 52926  3
mptscsih               36748  1 mptsas
mptbase                93903  2 mptsas,mptscsih
scsi_transport_sas     35086  1 mptsas

 

[marotori]

mptsas is an early megaraid card. Guess you are on a dell box?

 

[bazzi]

Nope on a Intel Modular Server

 

[tom]

latest is pve-kernel-2.6.32-11-pve: 2.6.32-65

 

[marotori]

What has changed with that version?

Rob

 

[tom]

see http://pve.proxmox.com/wiki/Roadmap#Proxmox_VE_2.0

 

[marotori]

Right...

Attached here:

View attachment hcpdump.zip

Is a script called 'hcpdump'

What this script does is to provide you with a similar but limited functionalty to vzdump. End result.. files can be restored by vsrestore and qmrestore.

So...

hcpdump all -> backs up all vms
hcpdump 100 -> backs up machine 100

Few things..

put the hcpdump.conf into /etc/ and adjust to suit.

put the hcpdump somewhere like /usr/local/bin/

Important to understand a few things. This will only backup machines stored in /var/lib/vz/

This is how my environment works - so have not mad it any fancier

Rob

 

[udi]

And everyone tested with the latest kernel?

i did not because it hurts when it hangs

and i saw people in the forum who had similar issues with newest kernels

 

[bazzi]

I just upgraded. I get it ruffly once a week, so I keep you posted!

 

[tux]

marotori really nice script!

I have reproduce the hangs and solve the issue. Please try again some changes.

#hcp -c /dev/pve/root /dev/pve/data

so use the -c paramter for your script. After using this parameter I do not have hangs angain.

# hcp /dev/pve/data
# hcp /dev/pve/data
# hcp /dev/pve/data
wait 30 seconds
# hcp -l
(see how all works - normal size grow)

# hcp --remove /dev/hcp2
(see how size grows extremly)

be fast before it grows to big and system hang :
# hcp --remove /dev/hcp1
# hcp --remove /dev/hcp3

With the -c parameter the problem with multi-snapshot is solved. Can you reproduce the problem?

 

[udi]

that's true because you store the changed blocks on different volume.

but why do you need multiple snapshots simultaneously? or was it an accident only?

 

[tux]

It is no accident. We got some big systems where our data change rapid. So there are running several systems on one host system. If one admin need a snapshot and a backup is running the system gets instable and hang. My experience, everyhing is possible and I want to find a solution for very special situations.
I will post my script, if it works perfect.

 

[udi]

@marotori:

does the quota setting hor hcp work for you?