Thank you Dietmar.
However I just realized that the cluster config refers to the quorum disk by
label, not by device.
So maybe this wasn't my problem in the first place.
Here's the cluster config I'm trying to apply:
<?xml version="1.0"?>
<cluster name="sthlm-bb" config_version="3">
<cman expected_votes="3" keyfile="/var/lib/pve-cluster/corosync.authkey"/>
<quorumd votes="1" allow_kill="0" interval="1" label="proxmox1_qdisk"
tko="10"/>
<totem token="54000"/>
<clusternodes>
<clusternode name="beast" votes="1" nodeid="1"/>
<clusternode name="beauty" votes="1" nodeid="2"/></clusternodes>
</cluster>
Now I'm not really sure what the status is of my cluster. I'm stuck at the line
on the wiki that says
"
Verify that rgmanager is running, if not, start it"
/etc/init.d/rgemanager start
doesn't work:
root@beast:/var/log# /etc/init.d/rgmanager status
rgmanager is stopped
My /var/log/daemon.log is filled with
Oct 2 10:55:11 beast pmxcfs[635625]: [status] crit: cpg_send_message failed: 9
root@beast:/var/log# clustat
Cluster Status for sthlm-bb @ Wed Oct 2 10:58:32 2013
Member Status: Quorate
Member Name ID Status
------ ---- ---- ------
beast 1 Online,
Local
beauty 2 Online
/dev/block/8:65 0 Online,
Quorum Disk
root@beauty:~# clustat
Cluster Status for sthlm-bb @ Wed Oct 2 10:58:50 2013
Member Status: Quorate
Member Name ID Status
------ ---- ---- ------
beast 1 Online
beauty 2 Online, Local
/dev/block/8:49 0 Online,
Quorum Disk
root@beast:/var/log# pvecm s
Version: 6.2.0
Config Version: 3
Cluster Name: sthlm-bb
Cluster Id: 28546
Cluster Member: Yes
Cluster Generation: 328
Membership state: Cluster-Member
Nodes: 2
Expected votes: 3
Quorum device votes: 1
Total votes: 3
Node votes: 1
Quorum: 2
Active subsystems: 7
Flags:
Ports Bound: 0 178
Node name: beast
Node ID: 1
Multicast addresses: 239.192.111.241
Node addresses: X.Y.Z.46
root@beauty:~# pvecm s
Version: 6.2.0
Config Version: 3
Cluster Name: sthlm-bb
Cluster Id: 28546
Cluster Member: Yes
Cluster Generation: 328
Membership state: Cluster-Member
Nodes: 2
Expected votes: 3
Quorum device votes: 1
Total votes: 3
Node votes: 1
Quorum: 2
Active subsystems: 7
Flags:
Ports Bound: 0 178
Node name: beauty
Node ID: 2
Multicast addresses: 239.192.111.241
Node addresses: X.Y.Z.47
My problem seems to be that rgmanager won't start, is this due to some
misconfig of the qdisk?
I can't find the error log from the rgmanager startup
Here's the info about the qdisk from both nodes:
root@beast:/var/log# mkqdisk -f proxmox1_qdisk
mkqdisk v1364188437
/dev/block/8:65:
/dev/disk/by-id/scsi-1494554000000000028d222db64a9650d05bebe3d5ddef499-part1:
/dev/disk/by-path/ip-10.6.6.21:3260-iscsi-iqn.2013-09.se.accumulate.sthlm:qrmdisk1-lun-0-part1:
/dev/sde1:
Magic: eb7a62c2
Label: proxmox1_qdisk
Created: Tue Oct 1 16:36:24 2013
Host: beast
Kernel Sector Size: 512
Recorded Sector Size: 512
root@beauty:~# mkqdisk -f proxmox1_qdisk
mkqdisk v1364188437
/dev/block/8:49:
/dev/disk/by-id/scsi-1494554000000000028d222db64a9650d05bebe3d5ddef499-part1:
/dev/disk/by-path/ip-10.6.6.21:3260-iscsi-iqn.2013-09.se.accumulate.sthlm:qrmdisk1-lun-0-part1:
/dev/sdd1:
Magic: eb7a62c2
Label: proxmox1_qdisk
Created: Tue Oct 1 16:36:24 2013
Host: beast
Kernel Sector Size: 512
Recorded Sector Size: 512
Any idea what I have done wrong?
Thanks,
Johan
On Oct 2, 2013, at 10:30 , Dietmar Maurer <[email protected]> wrote:
You can use the /dev/disk/by-XYZ device links.
From: [email protected]
[mailto:[email protected]] On Behalf Of Johan Wikström
Sent: Mittwoch, 02. Oktober 2013 10:18
To: [email protected]
Subject: [PVE-User] Quorom disk on different devices
Hi list!
I have set up a two node Proxmox cluster on two Dell servers with DRBD. Works
well, but I just realized
that I need a quorum disk for the HA setup. I created an iSCSI qdisk as per the
FAQ
(http://pve.proxmox.com/wiki/Two-Node_High_Availability_Cluster#Configuring_Fencing)
but I ran into a problem. For some reason the first server had one extra disk
device (floppy, maybe virtual floppy/USB) so now
the iSCSI disk is known as /dev/sde on the first server and /dev/sdd on the
second server.
Since the cluster config file is global I guess this will be a problem when
defining the quorum disk?
Do I need to start creating udev rules for the iSCSI LUN to persist the naming
on both hosts, or
can I cheat in some way? Maybe it would be enough to remove the "floppy" device
link /dev/sdd on the first
host and try to re-create the iSCSI target?
PVE Version:
pve-manager/3.1-3/dc0e9b0e (running kernel: 2.6.32-22-pve)
Hardware: 2 * Dell T620
Best regards
Johan Wikström
_______________________________________________
pve-user mailing list
[email protected]
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
