Dejan, The pid is now looking in the right place
pid=[/var/run/mysqld/mysqld.pid] Not sure why its still failing as I thought we'd fixed it when you pointed out the /var/run/mysql directory. I did a test with the older version of the OCF I had installed from 2.1.2 in place and it started fine with the rest of heartbeat being 2.1.3. so its definately something in the new OCF script. Also tried starting mysql as an lsb resource and that was successful. Thanks, Amy -----Original Message----- From: Dejan Muhamedagic <[EMAIL PROTECTED]> Sent: Thursday, February 14, 2008 1:20pm To: [EMAIL PROTECTED] Cc: [email protected] Subject: Re: FW: Re: FW: Re: [Linux-HA] mysql OCF issue on CentOS 5.1 after upgrading to heartbeat 2.1.3-3 Hi, On Thu, Feb 14, 2008 at 12:59:16PM -0500, [EMAIL PROTECTED] wrote: > Dejan, > > Here's the /var/log/messages section > > Feb 14 12:56:59 utnode1 lrmd: [7057]: WARN: mysql_repl:start process (PID > 7418) timed out (try 1). Killing with signal SIGTERM (15). > Feb 14 12:56:59 utnode1 lrmd: [7057]: WARN: operation start[16] on > ocf::mysql::mysql_repl for client 7060, its parameters: target_role=[started] > CRM_meta_timeout=[20000] crm_feature_set=[1.0.9] > pid=[/var/run/mysqld/mysqld.pid] datadir=[/replicated/mysql/data] : pid > [7418] timed out > Feb 14 12:56:59 utnode1 crmd: [7060]: ERROR: process_lrm_event: LRM operation > mysql_repl_start_0 (16) Timed Out (timeout=20000ms) > Feb 14 12:57:01 utnode1 crmd: [7060]: info: do_lrm_rsc_op: Performing > op=mysql_repl_stop_0 key=1:39:252588e5-3f75-4f81-a4f0-df0d4a643d0b) > Feb 14 12:57:01 utnode1 lrmd: [7057]: info: rsc:mysql_repl: stop > Feb 14 12:57:01 utnode1 lrmd: [7057]: info: RA output: > (mysql_repl:stop:stdout) OCF_RESKEY_pid is set to /var/run/mysqld/mysqld.pid This is what it should be set to, right? Then the RA fails and loops here: if [ ! -e $OCF_RESKEY_pid ]; then ocf_log debug "MySQL is not running" return $OCF_NOT_RUNNING; fi Sorry, I don't understand anymore. Where is the pid file actually stored? Or is that here, for some reason, the value of OCF_RESKEY_pid is wrong? Thanks, Dejan > Feb 14 12:57:01 utnode1 crmd: [7060]: info: process_lrm_event: LRM operation > mysql_repl_stop_0 (call=17, rc=0) complete > Feb 14 12:57:03 utnode1 crmd: [7060]: info: do_lrm_rsc_op: Performing > op=sendmail_stop_0 key=16:39:252588e5-3f75-4f81-a4f0-df0d4a643d0b) > > Thanks, > > amy > > -----Original Message----- > From: Dejan Muhamedagic <[EMAIL PROTECTED]> > Sent: Thursday, February 14, 2008 12:39pm > To: [EMAIL PROTECTED] > Cc: [email protected] > Subject: Re: FW: Re: FW: Re: [Linux-HA] mysql OCF issue on CentOS 5.1 after > upgrading to heartbeat 2.1.3-3 > > Hi, > > On Thu, Feb 14, 2008 at 11:47:48AM -0500, [EMAIL PROTECTED] wrote: > > Dejan, > > > > Ran it with this: > > > > # export OCF_ROOT=/usr/lib/ocf > > # OCF_RESKEY_datadir="/replicated/mysql/data" > > OCF_RESKEY_pid="/var/run/mysqld/mysqld.pid" sh -x > > /usr/lib/ocf/resource.d/heartbeat/mysql start > > Ok. Tried to read the code and it should be working, but as you > tried it... I have nowhere to test it. Could you add after line > 88 which says > > : ${OCF_RESKEY_pid=${OCF_RESKEY_pid_default}} > > just this echo command to make sure what's going on: > > echo OCF_RESKEY_pid is set to $OCF_RESKEY_pid > > Thanks, > > Dejan > > > > > Thanks, > > > > Amy > > > > -----Original Message----- > > From: Dejan Muhamedagic <[EMAIL PROTECTED]> > > Sent: Thursday, February 14, 2008 11:38am > > To: [EMAIL PROTECTED], [email protected] > > Subject: Re: FW: Re: FW: Re: [Linux-HA] mysql OCF issue on CentOS 5.1 after > > upgrading to heartbeat 2.1.3-3 > > > > Hi, > > > > This somehow drifted off the list. Let's try to keep it there. > > > > On Thu, Feb 14, 2008 at 10:18:32AM -0500, [EMAIL PROTECTED] wrote: > > > Ok more info. > > > > > > I added the pid parameter through the gui and it does show in > > > /var/lib/heartbeat/crm/cib.xml: > > > > > > <primitive class="ocf" type="mysql" provider="heartbeat" > > > id="mysql_repl"> > > > <instance_attributes id="mysql_repl_instance_attrs"> > > > <attributes> > > > <nvpair name="target_role" id="mysql_repl_target_role" > > > value="started"/> > > > <nvpair id="de970238-7dfb-4505-b623-5f316b3432454" > > > name="datadir" value="/replicated/mysql/data"/> > > > <nvpair id="b5dd43ba-297c-493a-999f-85ceb5f3f72f" > > > name="pid" value="/var/run/mysqld/mysqld.pid"/> > > > </attributes> > > > </instance_attributes> > > > </primitive> > > > > > > However when running by hand still getting this: > > > > How did you run it by hand? Did you add the new OCF_RESKEY_pid > > parameter too? That should work. > > > > > Looking at the 2.1.2 OCF it used to have this: > > > > > > : ${OCF_RESKEY_pid="/var/run/mysqld/mysqld.pid"} > > > > > > The 2.1.3 OCF now has this: > > > > > > OCF_RESKEY_pid_default="/var/mysql/mysqld.pid" > > > > That's actually for OpenBSD. But still I can see that the default > > was changed to > > > > OCF_RESKEY_pid_default="/var/run/mysql/mysqld.pid" > > > > which is a regression. > > > > > While it shouldn't be a problem to just set the parameter it > > > does make it impossible to run 2 different versions of the > > > heartbeat at the same time while doing a gradual upgrade (ie > > > one machine in production and one machine out of production > > > getting upgraded). > > > > > > Should this be reported as a bug? While I can work around it it > > > does mean potential production downtime while switching servers > > > around. > > > > Sorry about that. Perhaps it should be reverted back to what has > > been there before. A bugzilla entry would be good too. > > > > Thanks, > > > > Dejan > > > > > > > > Thanks, > > > > > > Amy > > > > > > > > > > > > -----Original Message----- > > > From: Dejan Muhamedagic <[EMAIL PROTECTED]> > > > Sent: Thursday, February 14, 2008 7:05am > > > To: [EMAIL PROTECTED] > > > Subject: Re: FW: Re: [Linux-HA] mysql OCF issue on CentOS 5.1 after > > > upgrading to heartbeat 2.1.3-3 > > > > > > Hi, > > > > > > On Wed, Feb 13, 2008 at 02:54:38PM -0500, [EMAIL PROTECTED] wrote: > > > > Dejan, > > > > > > > > Just a thought after reading other posts. Could the pacemaker > > > > packages be needed as their functionality has been seperated > > > > out? > > > > > > Not with the Heartbeat release 2.1.3. That release includes the > > > CRM (pacemaker). > > > > > > > passwd and group both in the stard location /etc > > > > > > > > nssswitch.conf > > > > passwd: files > > > > shadow: files > > > > group: files > > > > > > OK, this rules out a network delay. > > > > > > > Here is the output from the command line before I end it after getting > > > > stuck at the same place: > > > > > > > > # OCF_RESKEY_datadir="/replicated/mysql/data" sh -x > > > > /usr/lib/ocf/resource.d/heartbeat/mysql start > > > > + . /usr/lib/ocf/resource.d/heartbeat/.ocf-shellfuncs > > > [snip] > > > > + /sbin/restorecon /var/log/mysqld.log > > > > + '[' 0 -a '!' -d /replicated/mysql/data/mysql ']' > > > > + /usr/bin/safe_mysqld --defaults-file=/etc/my.cnf > > > > --pid-file=/var/run/mysql/mysqld.pid --socket=/var/lib/mysql/mysql.sock > > > > --datadir=/replicated/mysql/data --user=mysql > > > > + rc=0 > > > > + '[' 0 '!=' 0 ']' > > > > > > It looks as if mysql started here... > > > > > > > + start_wait=1 > > > > + '[' 1 = 1 ']' > > > > + mysql_status > > > > + '[' '!' -e /var/run/mysql/mysqld.pid ']' > > > > + ocf_log debug 'MySQL is not running' > > > > > > but the status reports that it is not because it can't find the > > > pid file. Please check where's the pid file and if mysql is > > > running or not. This looks like a mysql setup problem. > > > > > > Thanks, > > > > > > Dejan > > > > > > > > > > > > > > > _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
