[Pacemaker] "stonith_admin -F node" results in a pair of reboots
Digimer
lists at alteeve.ca
Wed Jan 1 06:04:57 UTC 2014
Did you hook DRBD into pacemaker's fencing using 'crm-fence-peer.sh' and
set the fencing policy to 'resource-and-stonith;'? If not, do so! It
will protect against split-brains.
digimer
On 01/01/14 01:03 AM, Bob Haxo wrote:
> Digimer,
>
> Ok, sounds reasonable and I will investigate this further on Jan 2. WRT
> DRBD ... geeee, I don't recall multiple fencings. I'll check that also
> on Jan 2.
>
> Emmanuel,
>
> I have not seen pending fencing operations with "dlm_tool ls" ... but I
> have seen the word "pending" elsewhere (crm_mon?) without considering
> that it might be fencing that is pending. Interesting.
>
> Thanks & my best wishes for a healthy new year.
> Bob Haxo
>
>
> On Wed, 2014-01-01 at 00:19 -0500, Digimer wrote:
>> This is probably because cman (which is it's own cluster stack and used
>> to provide DLM and quorum to pacemaker on EL6) detected the node failed
>> after the initial fence and called it's own fence. You see a similar
>> behaviour when using DRBD. It will also call a fence when the peer dies
>> (even when it died because of a controlled fence call). In theory,
>> pacemaker using cman's dlm with DRBD would trigger three fences per
>> failure. :)
>>
>> digimer
>>
>> On 01/01/14 12:04 AM, emmanuel segura wrote:
>> > maybe you missing log when you had fenced the node? because i think the
>> > clvmd hungup because your node are in unclean state, use dlm_tool ls to
>> > see if you any pending fencing operation.
>> >
>> >
>> > 2014/1/1 Bob Haxo <bhaxo at sgi.com <mailto:bhaxo at sgi.com> <mailto:bhaxo at sgi.com>>
>> >
>> > __
>> > Greetings ... Happy New Year!
>> >
>> > I am testing a configuration that is created from example in
>> > "Chapter 6. Configuring a GFS2 File System in a Cluster" of the "Red
>> > Hat Enterprise Linux 7.0 Beta Global File System 2" document. Only
>> > addition is stonith:fence_ipmilan. After encountering this issue
>> > when I configured with "crm", I re-configured using "pcs". I've
>> > included the configuration below.
>> >
>> > I'm thinking that, in a 2-node cluster, if I run "stonith_admin -F
>> > <peer-node>", then <peer-node> should reboot and cleanly rejoin the
>> > cluster. This is not happening.
>> >
>> > What ultimately happens is that after the initially fenced node
>> > reboots, the system from which the stonith_admin -F command was run
>> > is fenced and reboots. The fencing stops there, leaving the cluster
>> > in an appropriate state.
>> >
>> > The issue seems to reside with clvmd/lvm. With the reboot of the
>> > initially fenced node, the clvmd resource fails on the surviving
>> > node, with a maximum of errors. I hypothesize there is an issue
>> > with locks, but have insufficient knowledge of clvmd/lvm locks to
>> > prove or disprove this hypothesis.
>> >
>> > Have I missed something ...
>> >
>> > 1) Is this expected behavior, and always the reboot of the fencing
>> > node happens?
>> >
>> > 2) Or, maybe I didn't correctly duplicate the Chapter 6 example?
>> >
>> > 3) Or, perhaps something is wrong or omitted from the Chapter 6 example?
>> >
>> > Suggestions will be much appreciated.
>> >
>> > Thanks,
>> > Bob Haxo
>> >
>> > RHEL6.5
>> > pacemaker-cli-1.1.10-14.el6_5.1.x86_64
>> > crmsh-1.2.5-55.1sgi709r3.rhel6.x86_64
>> > pacemaker-libs-1.1.10-14.el6_5.1.x86_64
>> > cman-3.0.12.1-59.el6_5.1.x86_64
>> > pacemaker-1.1.10-14.el6_5.1.x86_64
>> > corosynclib-1.4.1-17.el6.x86_64
>> > corosync-1.4.1-17.el6.x86_64
>> > pacemaker-cluster-libs-1.1.10-14.el6_5.1.x86_64
>> >
>> > Cluster Name: mici
>> > Corosync Nodes:
>> >
>> > Pacemaker Nodes:
>> > mici-admin mici-admin2
>> >
>> > Resources:
>> > Clone: clusterfs-clone
>> > Meta Attrs: interleave=true target-role=Started
>> > Resource: clusterfs (class=ocf provider=heartbeat type=Filesystem)
>> > Attributes: device=/dev/vgha2/lv_clust2 directory=/images
>> > fstype=gfs2 options=defaults,noatime,nodiratime
>> > Operations: monitor on-fail=fence interval=30s
>> > (clusterfs-monitor-interval-30s)
>> > Clone: clvmd-clone
>> > Meta Attrs: interleave=true ordered=true target-role=Started
>> > Resource: clvmd (class=lsb type=clvmd)
>> > Operations: monitor on-fail=fence interval=30s
>> > (clvmd-monitor-interval-30s)
>> > Clone: dlm-clone
>> > Meta Attrs: interleave=true ordered=true
>> > Resource: dlm (class=ocf provider=pacemaker type=controld)
>> > Operations: monitor on-fail=fence interval=30s
>> > (dlm-monitor-interval-30s)
>> >
>> > Stonith Devices:
>> > Resource: p_ipmi_fencing_1 (class=stonith type=fence_ipmilan)
>> > Attributes: ipaddr=128.##.##.78 login=XXXXX passwd=XXXXX
>> > lanplus=1 action=reboot pcmk_host_check=static-list
>> > pcmk_host_list=mici-admin
>> > Meta Attrs: target-role=Started
>> > Operations: monitor start-delay=30 interval=60s timeout=30
>> > (p_ipmi_fencing_1-monitor-60s)
>> > Resource: p_ipmi_fencing_2 (class=stonith type=fence_ipmilan)
>> > Attributes: ipaddr=128.##.##.220 login=XXXXX passwd=XXXXX
>> > lanplus=1 action=reboot pcmk_host_check=static-list
>> > pcmk_host_list=mici-admin2
>> > Meta Attrs: target-role=Started
>> > Operations: monitor start-delay=30 interval=60s timeout=30
>> > (p_ipmi_fencing_2-monitor-60s)
>> > Fencing Levels:
>> >
>> > Location Constraints:
>> > Resource: p_ipmi_fencing_1
>> > Disabled on: mici-admin (score:-INFINITY)
>> > (id:location-p_ipmi_fencing_1-mici-admin--INFINITY)
>> > Resource: p_ipmi_fencing_2
>> > Disabled on: mici-admin2 (score:-INFINITY)
>> > (id:location-p_ipmi_fencing_2-mici-admin2--INFINITY)
>> > Ordering Constraints:
>> > start dlm-clone then start clvmd-clone (Mandatory)
>> > (id:order-dlm-clone-clvmd-clone-mandatory)
>> > start clvmd-clone then start clusterfs-clone (Mandatory)
>> > (id:order-clvmd-clone-clusterfs-clone-mandatory)
>> > Colocation Constraints:
>> > clusterfs-clone with clvmd-clone (INFINITY)
>> > (id:colocation-clusterfs-clone-clvmd-clone-INFINITY)
>> > clvmd-clone with dlm-clone (INFINITY)
>> > (id:colocation-clvmd-clone-dlm-clone-INFINITY)
>> >
>> > Cluster Properties:
>> > cluster-infrastructure: cman
>> > dc-version: 1.1.10-14.el6_5.1-368c726
>> > last-lrm-refresh: 1388530552
>> > no-quorum-policy: ignore
>> > stonith-enabled: true
>> > Node Attributes:
>> > mici-admin: standby=off
>> > mici-admin2: standby=off
>> >
>> >
>> > Last updated: Tue Dec 31 17:15:55 2013
>> > Last change: Tue Dec 31 16:57:37 2013 via cibadmin on mici-admin
>> > Stack: cman
>> > Current DC: mici-admin2 - partition with quorum
>> > Version: 1.1.10-14.el6_5.1-368c726
>> > 2 Nodes configured
>> > 8 Resources configured
>> >
>> > Online: [ mici-admin mici-admin2 ]
>> >
>> > Full list of resources:
>> >
>> > p_ipmi_fencing_1 (stonith:fence_ipmilan): Started
>> > mici-admin2
>> > p_ipmi_fencing_2 (stonith:fence_ipmilan): Started
>> > mici-admin
>> > Clone Set: clusterfs-clone [clusterfs]
>> > Started: [ mici-admin mici-admin2 ]
>> > Clone Set: clvmd-clone [clvmd]
>> > Started: [ mici-admin mici-admin2 ]
>> > Clone Set: dlm-clone [dlm]
>> > Started: [ mici-admin mici-admin2 ]
>> >
>> > Migration summary:
>> > * Node mici-admin:
>> > * Node mici-admin2:
>> >
>> > =====================================================
>> > crm_mon after the fenced node reboots. Shows the failure of clvmd
>> > that then
>> > occurs, which in turn triggers a fencing of that nnode
>> >
>> > Last updated: Tue Dec 31 17:06:55 2013
>> > Last change: Tue Dec 31 16:57:37 2013 via cibadmin on mici-admin
>> > Stack: cman
>> > Current DC: mici-admin - partition with quorum
>> > Version: 1.1.10-14.el6_5.1-368c726
>> > 2 Nodes configured
>> > 8 Resources configured
>> >
>> > Node mici-admin: UNCLEAN (online)
>> > Online: [ mici-admin2 ]
>> >
>> > Full list of resources:
>> >
>> > p_ipmi_fencing_1 (stonith:fence_ipmilan): Stopped
>> > p_ipmi_fencing_2 (stonith:fence_ipmilan): Started
>> > mici-admin
>> > Clone Set: clusterfs-clone [clusterfs]
>> > Started: [ mici-admin ]
>> > Stopped: [ mici-admin2 ]
>> > Clone Set: clvmd-clone [clvmd]
>> > clvmd (lsb:clvmd): FAILED mici-admin
>> > Stopped: [ mici-admin2 ]
>> > Clone Set: dlm-clone [dlm]
>> > Started: [ mici-admin mici-admin2 ]
>> >
>> > Migration summary:
>> > * Node mici-admin:
>> > clvmd: migration-threshold=1000000 fail-count=1
>> > last-failure='Tue Dec 31 17:04:29 2013'
>> > * Node mici-admin2:
>> >
>> > Failed actions:
>> > clvmd_monitor_30000 on mici-admin 'unknown error' (1): call=60,
>> > status=Timed Out, la
>> > st-rc-change='Tue Dec 31 17:04:29 2013', queued=0ms, exec=0ms
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> > _______________________________________________
>> > Pacemaker mailing list:Pacemaker at oss.clusterlabs.org <mailto:Pacemaker at oss.clusterlabs.org>
>> > <mailto:Pacemaker at oss.clusterlabs.org>
>> >http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>> >
>> > Project Home:http://www.clusterlabs.org
>> > Getting started:http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> > Bugs:http://bugs.clusterlabs.org
>> >
>> >
>> >
>> >
>> > --
>> > esta es mi vida e me la vivo hasta que dios quiera
>> >
>> >
>> > _______________________________________________
>> > Pacemaker mailing list:Pacemaker at oss.clusterlabs.org <mailto:Pacemaker at oss.clusterlabs.org>
>> >http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>> >
>> > Project Home:http://www.clusterlabs.org
>> > Getting started:http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> > Bugs:http://bugs.clusterlabs.org
>> >
>>
>>
>
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org
>
--
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?
More information about the Pacemaker
mailing list