Hi, I use a (externally developed) library within my application which used to work fine. However, after a recent update (which unfortunately we can't roll back due to dependencies) I have crazy CPU loads - there are ~20 cores in total on the webservers and they have gone from an average utilization of ~5% to >99%. I am using lighttpd, and so I can see the php-cgi processes using 99% all my CPU! I have an extremely basic understanding of how to track this down, and using XDebug and KCacheGrind I think that I have a pretty good idea which area the problem lies in, but the "problem" is that because whatever it is that has broken is using up the whole CPU everything else slows down. Furthermore, it would appear that a large part of what gets scheduled is luck because sometimes things are instant and the same activity a short while later can take forever. However, based on absolute CPU time's from this XDebug/KCacheGrind analysis and a gut hunch based upon what has changed recently I think I know roughly what it is. I do have a diff of the recent code change, and I have looked at it to no avail. My questions are: - Is there a better way to see what is using up the CPU? There must be a single loop or something talking to an external service, that is using so much CPU user time. Is there a super-quick way to do to this? I really need to know what function (or even line) is using all that CPU time, because I can't spot it in the code. Perhaps this information can be obtained from XDebug but I can't find it! - If there is no such magic answer to this question, what is the correct way to find such a phantom problem? Are there any other applications that might be better? If anyone has any personal recommendations for Consultants able to look at and fix this sort of problem that would be much appreciated too; I am now well out of my depth and sinking fast! Many thanks for any help, Alex -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php