I followed Clusters from Scratch instructions, chapters 8 and 9, to configure an active-active, shared disk cluster with drbd and pacemaker. The difference with that tutorial is that I am using OCFS2 instead of glusterfs.

Without pacemaker it works fine, but when I want to integrate to the cluster, it fails with this message of error:

# pcs status 
Cluster name: datacluster
Stack: corosync
Current DC: darwin (version 2.0.1-9e909a5bdd) - partition with quorum
Last updated: Wed May 26 00:18:41 2021
Last change: Tue May 25 23:54:38 2021 by root via cibadmin on darwin

2 nodes configured
3 resources configured

Online: [ darwin humboldt ]

Full list of resources:

 Clone Set: drbd-clone [drbd] (promotable)
     Masters: [ darwin humboldt ]
 shareddisk (ocf::heartbeat:Filesystem):    Stopped

Failed Resource Actions:
* shareddisk_start_0 on darwin 'not installed' (5): call=11, status=complete, exitreason='Couldn't find device [/dev/drbd0]. Expected /dev/??? to exist',
    last-rc-change='Wed May 26 00:02:45 2021', queued=0ms, exec=56ms
* shareddisk_start_0 on humboldt 'not installed' (5): call=12, status=complete, exitreason='Couldn't find device [/dev/drbd0]. Expected /dev/??? to exist',
    last-rc-change='Wed May 26 00:04:28 2021', queued=0ms, exec=56ms

Daemon Status:
  corosync: active/disabled
  pacemaker: active/disabled
  pcsd: active/enabled

I suspect that the message 'Couldn't find device [/dev/drbd0]. Expected /dev/??? to exist' is because the resource to mount the shared disk (shareddisk) doesn't wait until the resource drbd-clone has finished creating the device /dev/drbd0 in both nodes.

The device exist after the drbd resource is running.

root@darwin:~# cat /proc/drbd
version: 8.4.10 (api:1/proto:86-101)
srcversion: 15055BDD6F0D23278182874 
 0: cs:StandAlone ro:Primary/Unknown ds:UpToDate/DUnknown   r-----
    ns:0 nr:0 dw:0 dr:0 al:0 bm:0 lo:0 pe:0 ua:0 ap:0 ep:1 wo:f oos:0
root@darwin:~# ls -l /dev/drbd0
brw-rw---- 1 root disk 147, 0 May 26 00:02 /dev/drbd0

As you can see, the resource shareddisk is already constrained with colocation and order. Is there a way to tell the resource shareddisk to wait a minute after drbd before trying to start?

Or I should look elsewhere to solve this problem?

Thank you in advance for any help.

Here you can find both the cib and the configuration cib.

  • Can you share your cib so we can see how the resources are configured? Commented May 26, 2021 at 14:41
  • Sure. I just added the cib files to the question using pastebin. Commented May 26, 2021 at 15:16


You must log in to answer this question.

Browse other questions tagged .