Re: VM crash, store in glusterfs

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 07/02/2015 03:32 PM, Gregor Burck wrote:
Hi,

I would like to ade some test with glusterfs and virtualbox.
The goal is to store the virtualbox files in a glusterfs store and access it from to machines.

In my setup, the servers are the clients too. I hope you could understand my description:

I remount the share back to the single machines


two machines:
gf001 and gf002

Both:
debian 8
glusterfs 3.7

/dev/sda - root
/dev/sdb - /export/vbstore

My volume info:
root@gf001  :~# gluster volume info vbstore

Volume Name: vbstore
Type: Replicate
Volume ID: 7bf8aa42-8fd9-4535-888d-dacea4f14a83
Status: Started
Number of Bricks: 1 x 2 = 2
Transport-type: tcp
Bricks:
Brick1: gf001.mvz.ffm:/export/vbstore
Brick2: gf002.mvz.ffm:/export/vbstore
Options Reconfigured:
cluster.server-quorum-type: server
cluster.quorum-type: auto
network.remote-dio: enable
cluster.eager-lock: enable
performance.stat-prefetch: off
performance.io-cache: off
performance.read-ahead: off
performance.quick-read: off
performance.readdir-ahead: on


mounting on gf001:
mount -t glusterfs gf001:/vbstore /import/vbstore
or via fstab:
gf001:/vbstore        /import/vbstore glusterfs defaults 0 0

The same on gf002 with gf002 as server.

Creating an virtualbox VM in /import/vbstore.

Starting VM on gf001 OR gf002 work right.

But when running VM on one AND shut down the opposit, the VM crash with ATA failure.


Maybe mount or gluster setting options?

Thank you for help,

Gregor

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://www.gluster.org/mailman/listinfo/gluster-users
Hi Gregor,

Volume info shows that server-side quorum and client-side quorum are enabled.

If there are only 2 nodes ( gf001 & gf002 ) are there in the cluster, then
server-side quorum is not met, which would result in killing bricks of the volumes.
You can check this from 'gluster volume status <vol-name>'

Also client-side quorum has a limitation with replica 2 volume, that the first brick of the replicate should always be up. In your case, if you shutdown gf001, then the client-side quorum will not be met and
that would result in gluster volume becoming read-only

Under these situation, VM should not crash, but will go paused as the storage is inaccessible ( read-only in the case of client quorum not met )

Hope that helps.

Thanks,
Satheesaran S

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://www.gluster.org/mailman/listinfo/gluster-users



[Index of Archives]     [Gluster Development]     [Linux Filesytems Development]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux