On 1/9/07, Achim Stumpf <[EMAIL PROTECTED]> wrote:
Andrew Beekhof wrote:
> you might also want to check out the one under source control:
> http://hg.linux-ha.org/dev/file/tip/resources/OCF/mysql.in
Hi Andrew,
Thanks a lot for this script. I have installed it and tested it already,
but there was some stuff not working well.
Here is a diff with my changes:
--- mysql.new.ha-linux (revision 49)
+++ mysql.new.ha-linux (revision 48)
@@ -30,7 +30,7 @@
#######################################################################
# Initialization:
-. /usr/lib/heartbeat/ocf-shellfuncs
+. @hb_libdir@/ocf-shellfuncs
the build process would normally make this replacement for you
#######################################################################
@@ -45,7 +45,7 @@
: ${OCF_RESKEY_group="mysql"}
: ${OCF_RESKEY_log="/var/log/mysqld.log"}
-: ${OCF_RESKEY_pid="/var/run/mysqld/mysqld.pid"}
+: ${OCF_RESKEY_pid="/var/lib/mysql/mysqld.pid"}
: ${OCF_RESKEY_socket="/var/lib/mysql/mysql.sock"}
: ${OCF_RESKEY_test_user="root"}
@@ -207,7 +207,7 @@
kill -0 $pid >/dev/null 2>&1
fi
- if [ $? -eq 0 ]; then
+ if $?; then
Here I got before an error on the bash like:
bash: 0: command not found
return $OCF_SUCCESS;
else
ocf_log debug "MySQL not running: removing old PID file"
@@ -267,7 +267,7 @@
#chown -R $OCF_RESKEY_user $OCF_RESKEY_datadir
#chgrp -R $OCF_RESKEY_group $OCF_RESKEY_datadir
- $MYSQL_BINDIR/mysqld_safe --defaults-file=$OCF_RESKEY_config
--pid-file=$OCF_RESKEY_pid --socket=$OCF_RESKEY_socket
--datadir=$OCF_RESKEY_datadir --user=$OCF_RESKEY_user >/dev/null 2>&1 &
+ $MYSQL_BINDIR/safe_mysqld --defaults-file=$OCF_RESKEY_config
--pid-file=$OCF_RESKEY_pid --socket=$OCF_RESKEY_socket
--datadir=$OCF_RESKEY_datadir --user=$OCF_RESKEY_user >/dev/null 2>&1 &
rc=$?
Maybe it's better to put the mysql binary into a variable, because on
Fedora it's called the other way around.
done
So far the script was working then but there wasn't any monitoring log
messages in my logfiles of heartbeat. After a while I saw, that on line
223 a variable called $OCF_CHECK_LEVEL:
219 mysql_monitor() {
220 mysql_status
221 rc=$?
222
223 if [ $OCF_CHECK_LEVEL = 0 -o $rc != 0 ]; then
224 return $rc
225 fi
226
So in that case mysql_status finds the pidfile and returns a
$OCF_SUCCESS which is zero. So it should go on with the monitoring of
mysql, but the variable $OCF_CHECK_LEVEL is only defined on that line no
where else in the mysql script. I have found it also in ocf-shellfuncs,
but i don't have a clue what it does.
OCF_CHECK_LEVEL is part of the OCF standard.
you define it as a parameter to the monitor operation.
an example of setting extra parameters for individual operation can be
found at about line 31 of:
http://hg.linux-ha.org/dev/file/tip/crm/pengine/testcases/params-1.xml
instead of "new-attr", you'd specify OCF_CHECK_LEVEL
i've also made some of the other changes you recommended
So I took my script again... I know it works. But maybe you could
enlightened me a bit...
Thanks,
Achim
>
> On 1/8/07, Achim Stumpf <[EMAIL PROTECTED]> wrote:
>> Hi list,
>>
>> I have rewritten a mysql ocf script which I found on
>> http://k2k.ds14.agh.edu.pl/egee-ha/mysql
>>
>> Basically I corrected the exit codes and I have rewritten some functions
>> which i didn't like.
>>
>> In my test cluster it works fine. I haven't found any mysql ocf script
>> shipped with your source code, so maybe you like mine.
>>
>> If you have any advice, what should be changed or what else you need of
>> mine, so that you are able to integrate it in your sources, let me know
>> please.
>>
>>
>> cheers,
>>
>> Achim
>>
>>
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
