Confluence Questions lucene write.lock remain

[kunitaka]

Hi,

Now, I faced with performance problem about Confluence.

Once Confluence is slow, even it Confluence is restarted, problem is not resloved.

So, I delete disk cache of Linux, problem is resolved.

I don't know what is effect to disk cache.

Currently under investigation and I find strange somthing in Confluence Questions.

There is always a write.lock in directory.

<confluence home directory>/index/quetions-edge/

It is maked when confluence is started.

Maybe this file shuld be deleted. Is it right?

Confluence 5.7.1

Best Regards

[Nic Brough -Adaptavist-]

NO!  That file is there to prevent damage being done by trying to start other Confluence systems with the same data.  Do NOT touch it.  It is safe to delete it when Confluence is definitely stopped, but never never touch it (or any other Confluence files for that matter) while it is running.

I don't know what cache you've cleared (it's fine to clear most of them, again, while it's not running), but to be on the safe side, I would re-index Confluence through the admin UI, sometime when your users are not too active.

[kunitaka]

Thanks.

I refer to below

https://jira.atlassian.com/browse/CONF-29004

So, I think write.lock file of Confluence Questions should be deleted. 

And I have already done re-index, but problem is occured.

Many Thanks

[Nic Brough -Adaptavist-]

No.

Do not touch the file while Confluence is running.  Stop Confluence.  Remove the file.  Restart Confluence. 

Could you explain what the actual problem is here?  What symptoms are you investigating?

[kunitaka]

> Do not touch the file while Confluence is running.  Stop Confluence.  Remove the file.  Restart Confluence. 

Understand.

Could you explain what the actual problem is here?

Suddnely, CPU load , and DB(mysql) thread is up to limit.

This problem is sometimes occured. When it is occured, I restart confluence to reslove.

But, in this case, I could not reslove problem even if I restarted confluence.

So, to reslove, we tried other ways.

• MySQL restarted

And That I cleared disk cache is effectual.

echo 1 > /proc/sys/vm/drop_caches 

I suspect lucene is something wrong.  

[Nic Brough -Adaptavist-]

Ok, what makes you think it is Lucene?  Have you checked the log files and the cache settings inside Confluence?

[kunitaka]

I'm sorry. I'm late.

The cause is hardware trouble. A part of memory is not good.

Thank you for your  cooperation.