anson parker's picture

Hi all - my d7 site was running aces with mostly standard modules on a micro account until about 4pm this afternoon - for the past 8 hours it has been brutal.  like 100% CPU tacked out on basic request.  The site is in maintenance mode and there are no special modules - it's heinous.  Anyone got some tips on debugging or determining whether it's AWS, or me, or YOU (j/k turnkey is probably the only part of the system I'm confident in) or wt??

 

Thanks (and thanks again for turnkey, it rocks)

Forum: 
Tags: 
Jeremy Davis's picture

Assuming you're running a Micro instance that is...!

Once your appliance hits 100% CPU for more than about 5 seconds running you get throttled back to ~<5% CPU. What seems to happen then is that requests and other tasks back up, thus the appliance gets stuck. Everytime the throttling drops off, the CPU hits 100% straight away (to process outstanding stuff that has been backing up) so it gets throttled again...

So, solutions? Possibly the easiset (and maybe even best - depending on how much usage you expect your server to get) is to upgrade your server to a Small one. Another option is to look at AWS auto-scaling (I know nothing about it but another user mentioned it) - from what I can gather it'll launch additional load balancing Micro servers when required. The only other option that I am aware of would be to install something to thottle the CPU internally (so it never hits 100%). Something like this. Also, some time ago I posted about a daemon I found here (which I don't know much about, just found it and shared...) I have read of that approach being successful but haven't tested myself.

While this thread involves WordPress, it probably mostly applies to any Micro appliance.

So if it's not a Micro instance, then I apologise (and it's obviously not that...)

Add new comment