GFS2 Datastore mount problem

[YvesV]

Hell,

when trying to mount a GFS2 DataStore from a Alletra MP I'm getting this eror : 

Unable to create / update pcs stonith: Error: Unable to perform restartless update of scsi devices: resource 'hpevm_gfs2_scsi' is not running on any node, please use command 'pcs stonith update' instead

how can this be solved?

[PLalonde]

Hello,

We are having the same issue, but with an HPE MSA 2072. Ubuntu discovers the iSCSI luns properly across all hosts but VME won't allow us to create GFS2 filesystems.

[PeterTzvetanov]

Hi PLalonde,

so far I can see from the screenshot, there are only 2 hosts in that cluster. For the GFS2 datastore, you will need at least 3 ubuntu hosts in that cluster. With the 2 hosts, the only way for datastores is the NFS as there the min. requirement is at least 1 host.

With regards

[PLalonde]

Thank you for the follow-up.

There are in fact three hosts, but only two of them were shown in that screenshot. The source of this error was an attempt to create a second GFS2 datastore on a second iSCSI LUN while the first GFS2 datastore creation operation was still in progress. The end result was that VME trashed the Pacemaker fencing configuration for the second GFS2 datastore and nothing I did could make it usable. I had to completely re-install all three hosts. Here are the logs from systemctl status pacemaker on each host:

? pacemaker.service - Pacemaker High Availability Cluster Manager
     Loaded: loaded (/usr/lib/systemd/system/pacemaker.service; enabled; preset: enabled)
     Active: active (running) since Thu 2025-11-20 07:47:47 EST; 5min ago
       Docs: man:pacemakerd
             https://clusterlabs.org/pacemaker/doc/
   Main PID: 2752 (pacemakerd)
      Tasks: 7
     Memory: 41.8M (peak: 60.2M)
        CPU: 3.590s
     CGroup: /system.slice/pacemaker.service
             +-2752 /usr/sbin/pacemakerd
             +-2753 /usr/lib/pacemaker/pacemaker-based
             +-2754 /usr/lib/pacemaker/pacemaker-fenced
             +-2755 /usr/lib/pacemaker/pacemaker-execd
             +-2756 /usr/lib/pacemaker/pacemaker-attrd
             +-2757 /usr/lib/pacemaker/pacemaker-schedulerd
             +-2758 /usr/lib/pacemaker/pacemaker-controld

Nov 20 07:51:07 vmehost1 pacemaker-controld[2758]:  error: Unfencing of vmehost3 by vmehost2 failed (Error) with exit status 1
Nov 20 07:51:07 vmehost1 pacemaker-fenced[2754]:  error: Operation 'on' [3439] targeting vmehost1 using hpevm_gfs2_scsi returned 1
Nov 20 07:51:07 vmehost1 pacemaker-fenced[2754]:  warning: hpevm_gfs2_scsi[3439] [ /usr/sbin/fence_scsi:268: SyntaxWarning: invalid escape sequence '\s' ]
Nov 20 07:51:07 vmehost1 pacemaker-fenced[2754]:  warning: hpevm_gfs2_scsi[3439] [   if not re.search(r"^" + dev + "\s+", out, flags=re.MULTILINE): ]
Nov 20 07:51:07 vmehost1 pacemaker-fenced[2754]:  warning: hpevm_gfs2_scsi[3439] [ 2025-11-20 07:51:07,842 ERROR: Failed: device "/dev/disk/by-uuid/e366418a-e0c>
Nov 20 07:51:07 vmehost1 pacemaker-fenced[2754]:  warning: hpevm_gfs2_scsi[3439] [  ]
Nov 20 07:51:07 vmehost1 pacemaker-fenced[2754]:  warning: hpevm_gfs2_scsi[3439] [ 2025-11-20 07:51:07,842 ERROR: Please use '-h' for usage ]
Nov 20 07:51:07 vmehost1 pacemaker-fenced[2754]:  warning: hpevm_gfs2_scsi[3439] [  ]
Nov 20 07:51:07 vmehost1 pacemaker-fenced[2754]:  notice: Operation 'on' targeting vmehost1 by vmehost2 for pacemaker-controld.2779@vmehost2: Error occurred (co>
Nov 20 07:51:07 vmehost1 pacemaker-controld[2758]:  error: Unfencing of vmehost1 by vmehost2 failed (Error) with exit status 1
~

? pacemaker.service - Pacemaker High Availability Cluster Manager
     Loaded: loaded (/usr/lib/systemd/system/pacemaker.service; enabled; preset: enabled)
     Active: active (running) since Thu 2025-11-20 07:47:34 EST; 7min ago
       Docs: man:pacemakerd
             https://clusterlabs.org/pacemaker/doc/
   Main PID: 2773 (pacemakerd)
      Tasks: 7
     Memory: 58.6M (peak: 76.8M)
        CPU: 4.803s
     CGroup: /system.slice/pacemaker.service
             +-2773 /usr/sbin/pacemakerd
             +-2774 /usr/lib/pacemaker/pacemaker-based
             +-2775 /usr/lib/pacemaker/pacemaker-fenced
             +-2776 /usr/lib/pacemaker/pacemaker-execd
             +-2777 /usr/lib/pacemaker/pacemaker-attrd
             +-2778 /usr/lib/pacemaker/pacemaker-schedulerd
             +-2779 /usr/lib/pacemaker/pacemaker-controld

Nov 20 07:51:07 vmehost2 pacemaker-controld[2779]:  notice: Fence operation 42 for vmehost3 failed: Agent returned error (aborting transition)
Nov 20 07:51:07 vmehost2 pacemaker-controld[2779]:  warning: Too many failures (13) to fence vmehost3, giving up
Nov 20 07:51:07 vmehost2 pacemaker-controld[2779]:  error: Unfencing of vmehost3 by vmehost2 failed (Error) with exit status 1
Nov 20 07:51:07 vmehost2 pacemaker-fenced[2775]:  notice: Couldn't find anyone to fence (on) vmehost1 using any device
Nov 20 07:51:07 vmehost2 pacemaker-fenced[2775]:  error: Operation 'on' targeting vmehost1 by vmehost2 for pacemaker-controld.2779@vmehost2: Error occurred (complete)
Nov 20 07:51:07 vmehost2 pacemaker-controld[2779]:  notice: Fence operation 41 for vmehost1 failed: Agent returned error (aborting transition)
Nov 20 07:51:07 vmehost2 pacemaker-controld[2779]:  warning: Too many failures (13) to fence vmehost1, giving up
Nov 20 07:51:07 vmehost2 pacemaker-controld[2779]:  error: Unfencing of vmehost1 by vmehost2 failed (Error) with exit status 1
Nov 20 07:51:07 vmehost2 pacemaker-controld[2779]:  notice: Transition 12 (Complete=3, Pending=0, Fired=0, Skipped=0, Incomplete=24, Source=/var/lib/pacemaker/pengine/p>
Nov 20 07:51:07 vmehost2 pacemaker-controld[2779]:  notice: State transition S_TRANSITION_ENGINE -> S_IDLE
telecom@vmehost2:~$

? pacemaker.service - Pacemaker High Availability Cluster Manager
     Loaded: loaded (/usr/lib/systemd/system/pacemaker.service; enabled; preset: enabled)
     Active: active (running) since Thu 2025-11-20 07:47:25 EST; 7min ago
       Docs: man:pacemakerd
             https://clusterlabs.org/pacemaker/doc/
   Main PID: 2743 (pacemakerd)
      Tasks: 7
     Memory: 42.0M (peak: 60.1M)
        CPU: 4.581s
     CGroup: /system.slice/pacemaker.service
             +-2743 /usr/sbin/pacemakerd
             +-2745 /usr/lib/pacemaker/pacemaker-based
             +-2746 /usr/lib/pacemaker/pacemaker-fenced
             +-2747 /usr/lib/pacemaker/pacemaker-execd
             +-2748 /usr/lib/pacemaker/pacemaker-attrd
             +-2749 /usr/lib/pacemaker/pacemaker-schedulerd
             +-2750 /usr/lib/pacemaker/pacemaker-controld

Nov 20 07:51:07 vmehost3 pacemaker-fenced[2746]:  warning: hpevm_gfs2_scsi[3176] [ /usr/sbin/fence_scsi:268: SyntaxWarning: invalid escape sequence '\s' ]
Nov 20 07:51:07 vmehost3 pacemaker-fenced[2746]:  warning: hpevm_gfs2_scsi[3176] [   if not re.search(r"^" + dev + "\s+", out, flags=re.MULTILINE): ]
Nov 20 07:51:07 vmehost3 pacemaker-fenced[2746]:  warning: hpevm_gfs2_scsi[3176] [ 2025-11-20 07:51:07,817 ERROR: Failed: device "/dev/disk/by-uuid/e366418a-e0c6-410b-83c7-6>
Nov 20 07:51:07 vmehost3 pacemaker-fenced[2746]:  warning: hpevm_gfs2_scsi[3176] [  ]
Nov 20 07:51:07 vmehost3 pacemaker-fenced[2746]:  warning: hpevm_gfs2_scsi[3176] [ 2025-11-20 07:51:07,817 ERROR: Please use '-h' for usage ]
Nov 20 07:51:07 vmehost3 pacemaker-fenced[2746]:  warning: hpevm_gfs2_scsi[3176] [  ]
Nov 20 07:51:07 vmehost3 pacemaker-fenced[2746]:  notice: Operation 'on' targeting vmehost3 by vmehost2 for pacemaker-controld.2779@vmehost2: Error occurred (complete)
Nov 20 07:51:07 vmehost3 pacemaker-controld[2750]:  error: Unfencing of vmehost3 by vmehost2 failed (Error) with exit status 1
Nov 20 07:51:07 vmehost3 pacemaker-fenced[2746]:  notice: Operation 'on' targeting vmehost1 by vmehost2 for pacemaker-controld.2779@vmehost2: Error occurred (complete)
Nov 20 07:51:07 vmehost3 pacemaker-controld[2750]:  error: Unfencing of vmehost1 by vmehost2 failed (Error) with exit status 1

It should be possible to create a second GFS2 datastore while the first one is being created, otherwise a warning and a wait measure would be appropriate.

Paul

[PeterTzvetanov]

is it now possible for an remote session, if yes send a maill to hpe-sw-trial-vmessentials@hpe.com and we will have a look.

[PLalonde]

Thank you for the offer of assistance, but we went ahead and re-installed Ubuntu / VME on all three affected hosts. Once we created the first GFS2 datastore, we allowed it to fully complete before attempting to create the second one, and that was successful.

Paul