[ClusterLabs] need some help with failing resources

Mon Dec 5 15:55:19 UTC 2016

On 12/05/2016 09:30 AM, Darko Gavrilovic wrote:
> On 12/5/2016 10:17 AM, Ken Gaillot wrote:
>> On 12/03/2016 05:19 AM, Darko Gavrilovic wrote:
>>> Here is the output for that resource.. edited
>>>
>>> primitive svc-mysql ocf:heartbeat:mysql \
>>>         params binary="/usr/bin/mysqld_safe" config="/etc/my.cnf"
>>> datadir="/var/lib/mysql" user="mysql" group="mysql"
>>> log="/var/log/mysqld.log" pid="/var/run/mysqld/mysqld.pid"
>>> socket="/var/lib/mysql/mysql.sock" test_table="***" test_user="***"
>>> test_passwd="****" \
>>>         op monitor interval="30s" timeout="60s" OCF_CHECK_LEVEL="5" \
>>>         op start interval="0" timeout="120s" \
>>>         op stop interval="0" timeout="120s" \
>>>         meta target-role="Started" migration-threshold="2"
>>>
>>> ...skipping
>>> order mysql-before-httpd inf: svc-mysql:start svc-httpd:start
>>> order mysql-before-ssh inf: svc-mysql:start svc-ssh:start
>>> property $id="cib-bootstrap-options" \
>>>         dc-version="1.0.6-f709c638237cdff7556cb6ab615f32826c0f8c06" \
>>>         cluster-infrastructure="openais" \
>>>         expected-quorum-votes="2" \
>>>         last-lrm-refresh="1480762389" \
>>>         no-quorum-policy="ignore" \
>>>         stonith-enabled="true" \
>>>         ms-drbd0="Master"
>>>
>>>
>>> dg
>>>
>>>
>>> On 12/3/2016 1:25 AM, Kostiantyn Ponomarenko wrote:
>>>> I assume that you are using crmsh.
>>>> If so, I suggest to post an output from "crm configure show" command
>>>> here.
>>>>
>>>> Thank you,
>>>> Kostia
>>>>
>>>> On Sat, Dec 3, 2016 at 5:54 AM, Darko Gavrilovic
>>>> <darko at chass.utoronto.ca <mailto:darko at chass.utoronto.ca>> wrote:
>>>>
>>>>     Hello, I have a two node cluster running that seems to be
>>>> failing to
>>>>     start resources.
>>>>
>>>>      Resource Group: services
>>>>          svc-mysql  (ocf::heartbeat:mysql) Stopped
>>>>          svc-httpd  (ocf::heartbeat:apache) Stopped
>>>>          svc-ssh    (lsb:sshd-virt) Stopped
>>>>          svc-tomcat6        (lsb:tomcat6) Stopped
>>>>          svc-plone  (lsb:plone) Stopped
>>>>          svc-bacula (lsb:bacula-fd-virt) Stopped
>>>>
>>>>     When I run crm resource start services the service group does not
>>>> start.
>>>>
>>>>     I also tried starting the first resource in the group.
>>>>     crm resource start svc-mysql
>>>>
>>>>     It does not start either.
>>>>
>>>>     The error I am seeing is:
>>>>     Dec  2 21:59:43  pengine: [25829]: WARN: native_color: Resource
>>>>     svc-mysql cannot run anywhere
>>>>     Dec  2 22:00:26  pengine: [25829]: WARN: native_color: Resource
>>>>     svc-mysql cannot run anywhere
>>
>> The most common reasons for the above message are:
>>
>> * Location or order constraints don't leave any place for the resource
>> to run
>>
>> * The resource has failed the maximum number of times on all nodes.
>> (Does "crm_mon" show any failures?)
> 
> crm_mon does not list any failures for this service group from what I
> can see.
> 
>>
>>>>
>>>>     4b4f-a239-8a10dad40587, cib=0.3857.2) : Resource op removal
>>>>     Dec  2 21:59:32 server1 crmd: [25830]: info: te_rsc_command:
> <snip>
> 
>>>>     Initiating action 56: monitor svc-mysql_monitor_0 on server2
>>>>     Dec  2 21:59:33 server1 crmd: [25830]: WARN: status_from_rc: Action
>>>>     56 (svc-mysql_monitor_0) on server2 failed (target: 7 vs. rc: 0):
>>>> Error
>>
>> The above error indicates that mysql is running on server2 but the
>> cluster didn't start it there. (The "_monitor_0" is called a "probe";
>> it's used to check the status of the service before the cluster starts
>> it. The "target: 7" means it expects the service to be stopped. The "rc:
>> 0" means the service is actually running.)
>>
>> Make sure you're not starting mysql at boot or by any other means than
>> the cluster.
> 
> Thanks. Is there a way or command for me to check on which server does
> the cluster think the service last started on?

crm_mon shows what the cluster thinks the current state is, including
what resources are started where.

For way more (user-unfriendly) detail, you can look at the operation
history in the CIB XML, which you can see with (for example) "cibadmin
-Q". The <status> section will have a <node_state> entry for each node,
with a <lrm_resource> entry for each resource, with <lrm_rsc_op> entries
for all failed operations and the most recent successful operation. So
you can see all the *_start_0 operations with a whole bunch of
information including rc-code which is what it returned (0 is success).

>>>>     Dec  2 21:59:33 server1 crmd: [25830]: info:
>>>> abort_transition_graph:
>>>>     match_graph_event:272 - Triggered transition abort (complete=0,
>>>>     tag=lrm_rsc_op, id=svc-mysql_monitor_0,
>>>>     magic=0:0;56:15:7:aee06ee3-9576-4b4f-a239-8a10dad40587,
>>>>     cib=0.3859.2) : Event failed
>>>>     Dec  2 21:59:33 server1 crmd: [25830]: info: match_graph_event:
>>>>     Action svc-mysql_monitor_0 (56) confirmed on server2 (rc=4)
>>>>     Dec  2 21:59:33 server1 crmd: [25830]: info: te_rsc_command:
>>>>     Initiating action 187: stop svc-mysql_stop_0 on server2
>>>>     Dec  2 21:59:35 server1 crmd: [25830]: info: match_graph_event:
>>>>     Action svc-mysql_stop_0 (187) confirmed on server2 (rc=0)
>>>>     Dec  2 22:10:20 server1 crmd: [19708]: info: do_lrm_rsc_op:
>>>>     Performing key=101:1:7:6e477ca6-4ffe-4e89-82c2-c6149d528128
>>>>     op=svc-mysql_monitor_0 )
>>>>     Dec  2 22:10:20 server1 crmd: [19708]: info: process_lrm_event: LRM
>>>>     operation svc-mysql_monitor_0 (call=51, rc=7, cib-update=42,
>>>>     confirmed=true) not running
>>>>     Dec  2 22:12:24 server1 crmd: [19708]: info: te_rsc_command:
>>>>     Initiating action 102: monitor svc-mysql_monitor_0 on server2
>>>>     Dec  2 22:12:24 server1 crmd: [19708]: info: match_graph_event:
>>>>     Action svc-mysql_monitor_0 (102) confirmed on server2 (rc=0)
>>>>
>>>>
>>>>     Any advice on how to tackle this?
>>>>
>>>>     dg