[Ocfs2-tools-devel] [PATCH 3/6] o2cb: don't write sysconfig if status not changed

[Junxiao Bi]

On 07/28/2015 01:28 PM, Zhen Ren wrote:
> Hi Junxiao,
> 
> Did you notice the following problematic scenarios with o2cb cluster stack?
> 
> if you want to mount ocfs2 at boot  time by adding record to /etc/fstab (any other better method can do this?), 

Besides this, maybe you can write a systemd mount config file.
after cluster bring up,
> each node success to mount, 'systemctl status o2cb' looks good, but 'systemctl status ocfs2' looks  bad. At this time, using 'systemctl
> stop ocfs2' won't work because it has been recorded 'stopped'.  Inconsistency happens.
Yes, i just found this through the test. Looks like ocfs2 init script is
not called by "systemctl status ocfs2", you can try "systemctl restart
ocfs2" to have an consistent state.

Thanks,
Junxiao.
> 
> How should we handle this situation?
> --
> s3:~ # systemctl status o2cb
> o2cb.service - LSB: Load O2CB cluster services at system boot.
>    Loaded: loaded (/etc/init.d/o2cb)
>    Active: active (running) since Mon 2015-07-27 10:37:29 CST; 1 day 2h ago
>   Process: 1259 ExecStart=/etc/init.d/o2cb start (code=exited, status=0/SUCCESS)
>    CGroup: /system.slice/o2cb.service
>            └─1365 o2hbmonitor
> 
> Jul 27 10:37:28 s3 o2cb[1259]: Loading filesystem "configfs": OK
> Jul 27 10:37:28 s3 o2cb[1259]: Loading stack plugin "o2cb": OK
> Jul 27 10:37:29 s3 o2cb[1259]: Loading filesystem "ocfs2_dlmfs": OK
> Jul 27 10:37:29 s3 o2cb[1259]: Mounting ocfs2_dlmfs filesystem at /dlm: OK
> Jul 27 10:37:29 s3 o2cb[1259]: Setting cluster stack "o2cb": OK
> Jul 27 10:37:29 s3 o2cb[1259]: Registering O2CB cluster "hacluster": OK
> Jul 27 10:37:29 s3 o2cb[1259]: Setting O2CB cluster timeouts : OK
> Jul 27 10:37:29 s3 o2hbmonitor[1365]: Starting
> 
> --
> s3:~ # systemctl status ocfs2
> ocfs2.service - LSB: Mount OCFS2 volumes at boot.
>    Loaded: loaded (/etc/init.d/ocfs2)
>    Active: failed (Result: exit-code) since Mon 2015-07-27 10:37:29 CST; 1 day 2h ago
>   Process: 1366 ExecStart=/etc/init.d/ocfs2 start (code=exited, status=1/FAILURE)
> 
> Jul 27 10:37:29 s3 ocfs2[1366]: Starting Oracle Cluster File System (OCFS2) mount: can't find UUID=251fa3d4-fe6a-4c73-97dd-43026e76c05c
> Jul 27 10:37:29 s3 ocfs2[1366]: ..failed
> 
>  >>>
>> Signed-off-by: Junxiao Bi <junxiao.bi at oracle.com> 
>> --- 
>>  vendor/common/o2cb.init.sh |   16 ++++++++++------ 
>>  1 file changed, 10 insertions(+), 6 deletions(-) 
>>  
>> diff --git a/vendor/common/o2cb.init.sh b/vendor/common/o2cb.init.sh 
>> index 9e0b31c..e2f80f7 100755 
>> --- a/vendor/common/o2cb.init.sh 
>> +++ b/vendor/common/o2cb.init.sh 
>> @@ -1925,16 +1925,20 @@ case "$1" in 
>>          ;; 
>>   
>>      enable) 
>> -        O2CB_ENABLED=true 
>> -        write_sysconfig 
>> -        if_fail "$?" "Unable to write the driver configuration" 
>> +        if [ "$O2CB_ENABLED" != "true" ]; then 
>> +            O2CB_ENABLED=true 
>> +            write_sysconfig 
>> +            if_fail "$?" "Unable to write the driver configuration" 
>> +        fi 
>>          start 
>>          ;; 
>>   
>>      disable) 
>> -        O2CB_ENABLED=false 
>> -        write_sysconfig 
>> -        if_fail "$?" "Unable to write the driver configuration" 
>> +        if [ "$O2CB_ENABLED" != "false" ]; then 
>> +            O2CB_ENABLED=false 
>> +            write_sysconfig 
>> +            if_fail "$?" "Unable to write the driver configuration" 
>> +        fi 
>>          stop 
>>          ;; 
>>   
>  
> 
> 
> 
> --
> Eric, Ren
> 
>