Jump to content

Daily crashes - advice needed on optimization?


Recommended Posts

Posted

I upgraded to 3.4.5 at the same time I UPGRADED to a VPS from shared. Has 768mb RAM.

Since then I have daily crashes - the host says:

I am seeing your VPS using is running out of memory, currently the DNS service crashed as it was not able to run with the available resources left on your system. I have tweaked a few Apache settings to allow for it to use slightly less RAM, once these settings were checked the system committed ram is now lower in percentage. Your SWAP memory has now dropped in usage as well, this will help a little with performance. As it's quite quiet today due to the holidays, it's hard to see what your VPS will be like with some load on it. I would however say, would you have any issues with me installing caching software on your VPS to try to bring down the memory usage on MySQL with queries, but if you forums are very active this would have a negative effect there where updates wouldn't show for 5-10 minutes with NGINX installed.

The forum probably has maximum 200-300 people active every hour. Around 2,500 visits a day, say 15,000 page views.

Any tips an tricks for tuning IPB and also anyone know if any hooks, mods etc. know to use a lot of memory?

Thanks,

Simon

Posted

I didn't do the ip file edit but did apply the tapatalk update and saw drastic drop in tables mentioned. I didn't have the amount others had but still saw a drop so expect it would help larger also.

on the OP have you installed APC or xcache? that may help.

hard saying where the issue is really. could be sql too, w/o specifics hard to say.

Posted

Go to your Admin CP, then try this:

System Settings -> Advanced -> Cpu Saving Optimization

Now look at "Mark topics a user has posted when displaying a forum" option. If its turned on, turn it off, and check if your board works 100% now.

Posted

I didn't do the ip file edit but did apply the tapatalk update and saw drastic drop in tables mentioned. I didn't have the amount others had but still saw a drop so expect it would help larger also.

on the OP have you installed APC or xcache? that may help.

hard saying where the issue is really. could be sql too, w/o specifics hard to say.

Yes the impact is mainly on large sites, most won't notice an issue, however as with most coding issues, you take a little bug, toss 1000 users at it in 15 minutes and BOOM :smile: also no need for that edit.

Posted

Go to your Admin CP, then try this:

System Settings -> Advanced -> Cpu Saving Optimization

Now look at "Mark topics a user has posted when displaying a forum" option. If its turned on, turn it off, and check if your board works 100% now.

I've done that bud, is it a know problem?

What do members lose on the front end when turned off?

Posted

I've done that bud, is it a know problem?

What do members lose on the front end when turned off?

When you enter a section, you see the topics with a circle or a star on the left side.

The star means the user has posts on that topic.

With that function turned off, you will always see a circle and not the star. The query to check if a user has made posts on that topics is very resource intensive, specially if you have many users at that time visiting your board.

Posted

I'd disable prefetching and disable topic marking for guests before doing that.

really messes with user experience IMO.

what http software (the OP says apache and nginx), mysql versions and php versions you on?

just not enough info given to really suggest anything.

Posted
PHP Version 5.3.23

myslql 5.1.70

I've toggles that setting back on for now, as i don't really want to switch useful features off.

Since talking to you earlier I have been checking some other settings and I have employed a more permanent fix for apache crashing, apache was crashing every 5 minutes, I have been working on yoru server along with my other workload today and have your server monitoring dedicated to a screen on my desk. I was able to catch them within seconds of issues happening, since the fix was applied, which was to reset all events in the server. For some reason it was getting stuck in a cycle of thinking there was an issue and ending httpd, that issue has not been resolved. I notice your memory is looking good as well Munin is still running on your server to catch any specific load issues, so far it all seems to be working fine. I will keep this up while I am on shift, along with this ticket. If I see problems I will try to get them as they arise. I am hoping there are no more. I will advise you. I just didn't get a chance to let you know about the issues I was working on as I was tied up in a few other tickets. Your cloud is definitely an upgrade to shared, I understand you want your sites to be 100% and we are working on getting everything tweaked. If it was just as simple as making your vps imitate the shared server it would be no problem. Unfortunately it's not that simple. Sometimes VPS's depending on the web software being run has to be manually tweaked, thats what were here for. I hope you understand that we are trying our best to make your sites work 100% of the time, lets hope the last fixes put in place are now the end of your issues. I'll keep in touch with you on this through this ticket. I will set it to On Hold. I do appreciate your patience on this.

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...