[Pacemaker] hangs pending

Andrey Groshev greenx at yandex.ru
Wed Jan 15 13:41:47 UTC 2014



15.01.2014, 02:53, "Andrew Beekhof" <andrew at beekhof.net>:
> On 15 Jan 2014, at 12:15 am, Andrey Groshev <greenx at yandex.ru> wrote:
>
>>  14.01.2014, 10:00, "Andrey Groshev" <greenx at yandex.ru>:
>>>  14.01.2014, 07:47, "Andrew Beekhof" <andrew at beekhof.net>:
>>>>   Ok, here's what happens:
>>>>
>>>>   1. node2 is lost
>>>>   2. fencing of node2 starts
>>>>   3. node2 reboots (and cluster starts)
>>>>   4. node2 returns to the membership
>>>>   5. node2 is marked as a cluster member
>>>>   6. DC tries to bring it into the cluster, but needs to cancel the active transition first.
>>>>      Which is a problem since the node2 fencing operation is part of that
>>>>   7. node2 is in a transition (pending) state until fencing passes or fails
>>>>   8a. fencing fails: transition completes and the node joins the cluster
>>>>
>>>>   Thats in theory, except we automatically try again. Which isn't appropriate.
>>>>   This should be relatively easy to fix.
>>>>
>>>>   8b. fencing passes: the node is incorrectly marked as offline
>>>>
>>>>   This I have no idea how to fix yet.
>>>>
>>>>   On another note, it doesn't look like this agent works at all.
>>>>   The node has been back online for a long time and the agent is still timing out after 10 minutes.
>>>>   So "Once the script makes sure that the victim will rebooted and again available via ssh - it exit with 0." does not seem true.
>>>  Damn. Looks like you're right. At some time I broke my agent and had not noticed it. Who will understand.
>>  I repaired my agent - after send reboot he is wait STDIN.
>>  Returned "normally" a behavior - hangs "pending", until manually send reboot. :)
>
> Right. Now you're in case 8b.
>
> Can you try this patch:  http://paste.fedoraproject.org/68450/38973966


Killed all day experiences.
It turns out here that:
1. Did cluster.
2. On the node-2 send signal (-4) - killed corosink
3. From node-1 (there DC) - stonith sent reboot
4. Noda rebooted and resources start.
5. Again. On the node-2 send signal (-4) - killed corosink
6. Again. From node-1 (there DC) - stonith sent reboot
7. Noda-2 rebooted and hangs in "pending"
8. Waiting, waiting..... manually reboot.
9. Noda-2 reboot and raised resources start.
10. GOTO p.2



>>  New logs: http://send2me.ru/crmrep1.tar.bz2
>>>>   On 14 Jan 2014, at 1:19 pm, Andrew Beekhof <andrew at beekhof.net> wrote:
>>>>>    Apart from anything else, your timeout needs to be bigger:
>>>>>
>>>>>    Jan 13 12:21:36 [17223] dev-cluster2-node1.unix.tensor.ru stonith-ng: (  commands.c:1321  )   error: log_operation: Operation 'reboot' [11331] (call 2 from crmd.17227) for host 'dev-cluster2-node2.unix.tensor.ru' with device 'st1' returned: -62 (Timer expired)
>>>>>
>>>>>    On 14 Jan 2014, at 7:18 am, Andrew Beekhof <andrew at beekhof.net> wrote:
>>>>>>    On 13 Jan 2014, at 8:31 pm, Andrey Groshev <greenx at yandex.ru> wrote:
>>>>>>>    13.01.2014, 02:51, "Andrew Beekhof" <andrew at beekhof.net>:
>>>>>>>>    On 10 Jan 2014, at 9:55 pm, Andrey Groshev <greenx at yandex.ru> wrote:
>>>>>>>>>    10.01.2014, 14:31, "Andrey Groshev" <greenx at yandex.ru>:
>>>>>>>>>>    10.01.2014, 14:01, "Andrew Beekhof" <andrew at beekhof.net>:
>>>>>>>>>>>    On 10 Jan 2014, at 5:03 pm, Andrey Groshev <greenx at yandex.ru> wrote:
>>>>>>>>>>>>     10.01.2014, 05:29, "Andrew Beekhof" <andrew at beekhof.net>:
>>>>>>>>>>>>>      On 9 Jan 2014, at 11:11 pm, Andrey Groshev <greenx at yandex.ru> wrote:
>>>>>>>>>>>>>>       08.01.2014, 06:22, "Andrew Beekhof" <andrew at beekhof.net>:
>>>>>>>>>>>>>>>       On 29 Nov 2013, at 7:17 pm, Andrey Groshev <greenx at yandex.ru> wrote:
>>>>>>>>>>>>>>>>        Hi, ALL.
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>        I'm still trying to cope with the fact that after the fence - node hangs in "pending".
>>>>>>>>>>>>>>>       Please define "pending".  Where did you see this?
>>>>>>>>>>>>>>       In crm_mon:
>>>>>>>>>>>>>>       ......
>>>>>>>>>>>>>>       Node dev-cluster2-node2 (172793105): pending
>>>>>>>>>>>>>>       ......
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>       The experiment was like this:
>>>>>>>>>>>>>>       Four nodes in cluster.
>>>>>>>>>>>>>>       On one of them kill corosync or pacemakerd (signal 4 or 6 oк 11).
>>>>>>>>>>>>>>       Thereafter, the remaining start it constantly reboot, under various pretexts, "softly whistling", "fly low", "not a cluster member!" ...
>>>>>>>>>>>>>>       Then in the log fell out "Too many failures ...."
>>>>>>>>>>>>>>       All this time in the status in crm_mon is "pending".
>>>>>>>>>>>>>>       Depending on the wind direction changed to "UNCLEAN"
>>>>>>>>>>>>>>       Much time has passed and I can not accurately describe the behavior...
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>       Now I am in the following state:
>>>>>>>>>>>>>>       I tried locate the problem. Came here with this.
>>>>>>>>>>>>>>       I set big value in property stonith-timeout="600s".
>>>>>>>>>>>>>>       And got the following behavior:
>>>>>>>>>>>>>>       1. pkill -4 corosync
>>>>>>>>>>>>>>       2. from node with DC call my fence agent "sshbykey"
>>>>>>>>>>>>>>       3. It sends reboot victim and waits until she comes to life again.
>>>>>>>>>>>>>      Hmmm.... what version of pacemaker?
>>>>>>>>>>>>>      This sounds like a timing issue that we fixed a while back
>>>>>>>>>>>>     Was a version 1.1.11 from December 3.
>>>>>>>>>>>>     Now try full update and retest.
>>>>>>>>>>>    That should be recent enough.  Can you create a crm_report the next time you reproduce?
>>>>>>>>>>    Of course yes. Little delay.... :)
>>>>>>>>>>
>>>>>>>>>>    ......
>>>>>>>>>>    cc1: warnings being treated as errors
>>>>>>>>>>    upstart.c: In function ‘upstart_job_property’:
>>>>>>>>>>    upstart.c:264: error: implicit declaration of function ‘g_variant_lookup_value’
>>>>>>>>>>    upstart.c:264: error: nested extern declaration of ‘g_variant_lookup_value’
>>>>>>>>>>    upstart.c:264: error: assignment makes pointer from integer without a cast
>>>>>>>>>>    gmake[2]: *** [libcrmservice_la-upstart.lo] Error 1
>>>>>>>>>>    gmake[2]: Leaving directory `/root/ha/pacemaker/lib/services'
>>>>>>>>>>    make[1]: *** [all-recursive] Error 1
>>>>>>>>>>    make[1]: Leaving directory `/root/ha/pacemaker/lib'
>>>>>>>>>>    make: *** [core] Error 1
>>>>>>>>>>
>>>>>>>>>>    I'm trying to solve this a problem.
>>>>>>>>>    Do not get solved quickly...
>>>>>>>>>
>>>>>>>>>    https://developer.gnome.org/glib/2.28/glib-GVariant.html#g-variant-lookup-value
>>>>>>>>>    g_variant_lookup_value () Since 2.28
>>>>>>>>>
>>>>>>>>>    # yum list installed glib2
>>>>>>>>>    Loaded plugins: fastestmirror, rhnplugin, security
>>>>>>>>>    This system is receiving updates from RHN Classic or Red Hat Satellite.
>>>>>>>>>    Loading mirror speeds from cached hostfile
>>>>>>>>>    Installed Packages
>>>>>>>>>    glib2.x86_64                                                              2.26.1-3.el6                                                               installed
>>>>>>>>>
>>>>>>>>>    # cat /etc/issue
>>>>>>>>>    CentOS release 6.5 (Final)
>>>>>>>>>    Kernel \r on an \m
>>>>>>>>    Can you try this patch?
>>>>>>>>    Upstart jobs wont work, but the code will compile
>>>>>>>>
>>>>>>>>    diff --git a/lib/services/upstart.c b/lib/services/upstart.c
>>>>>>>>    index 831e7cf..195c3a4 100644
>>>>>>>>    --- a/lib/services/upstart.c
>>>>>>>>    +++ b/lib/services/upstart.c
>>>>>>>>    @@ -231,12 +231,21 @@ upstart_job_exists(const char *name)
>>>>>>>>    static char *
>>>>>>>>    upstart_job_property(const char *obj, const gchar * iface, const char *name)
>>>>>>>>    {
>>>>>>>>    +    char *output = NULL;
>>>>>>>>    +
>>>>>>>>    +#if !GLIB_CHECK_VERSION(2,28,0)
>>>>>>>>    +    static bool err = TRUE;
>>>>>>>>    +
>>>>>>>>    +    if(err) {
>>>>>>>>    +        crm_err("This version of glib is too old to support upstart jobs");
>>>>>>>>    +        err = FALSE;
>>>>>>>>    +    }
>>>>>>>>    +#else
>>>>>>>>       GError *error = NULL;
>>>>>>>>       GDBusProxy *proxy;
>>>>>>>>       GVariant *asv = NULL;
>>>>>>>>       GVariant *value = NULL;
>>>>>>>>       GVariant *_ret = NULL;
>>>>>>>>    -    char *output = NULL;
>>>>>>>>
>>>>>>>>       crm_info("Calling GetAll on %s", obj);
>>>>>>>>       proxy = get_proxy(obj, BUS_PROPERTY_IFACE);
>>>>>>>>    @@ -272,6 +281,7 @@ upstart_job_property(const char *obj, const gchar * iface, const char *name)
>>>>>>>>
>>>>>>>>       g_object_unref(proxy);
>>>>>>>>       g_variant_unref(_ret);
>>>>>>>>    +#endif
>>>>>>>>       return output;
>>>>>>>>    }
>>>>>>>    Ok :) I patch source.
>>>>>>>    Type "make rc" - the same error.
>>>>>>    Because its not building your local changes
>>>>>>>    Make new copy via "fetch" - the same error.
>>>>>>>    It seems that if not exist ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3.tar.gz, then download it.
>>>>>>>    Otherwise use exist archive.
>>>>>>>    Cutted log .......
>>>>>>>
>>>>>>>    # make rc
>>>>>>>    make TAG=Pacemaker-1.1.11-rc3 rpm
>>>>>>>    make[1]: Entering directory `/root/ha/pacemaker'
>>>>>>>    rm -f pacemaker-dirty.tar.* pacemaker-tip.tar.* pacemaker-HEAD.tar.*
>>>>>>>    if [ ! -f ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3.tar.gz ]; then                                             \
>>>>>>>             rm -f pacemaker.tar.*;                                              \
>>>>>>>             if [ Pacemaker-1.1.11-rc3 = dirty ]; then                                   \
>>>>>>>                 git commit -m "DO-NOT-PUSH" -a;                                 \
>>>>>>>                 git archive --prefix=ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3/ HEAD | gzip > ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3.tar.gz;       \
>>>>>>>                 git reset --mixed HEAD^;                                        \
>>>>>>>             else                                                                \
>>>>>>>                 git archive --prefix=ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3/ Pacemaker-1.1.11-rc3 | gzip > ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3.tar.gz;     \
>>>>>>>             fi;                                                                 \
>>>>>>>             echo `date`: Rebuilt ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3.tar.gz;                                     \
>>>>>>>         else                                                                    \
>>>>>>>             echo `date`: Using existing tarball: ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3.tar.gz;                     \
>>>>>>>         fi
>>>>>>>    Mon Jan 13 13:23:21 MSK 2014: Using existing tarball: ClusterLabs-pacemaker-Pacemaker-1.1.11-rc3.tar.gz
>>>>>>>    .......
>>>>>>>
>>>>>>>    Well, "make rpm" - build rpms and I create cluster.
>>>>>>>    I spent the same tests and confirmed the behavior.
>>>>>>>    crm_reoprt log here - http://send2me.ru/crmrep.tar.bz2
>>>>>>    Thanks!
>>>>   ,
>>>>   _______________________________________________
>>>>   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
>>>  _______________________________________________
>>>  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
>>  _______________________________________________
>>  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
>
> ,
> _______________________________________________
> 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




More information about the Pacemaker mailing list