Hello All, Not sure if this is a common known tuning/configuration issue or not but I could not find any direct hits searching online for what could be causing this. I have an Apache server configured *purely* as a reverse proxy for a few apps. One of the recent apps added behind it has some really slow responding pages which may contribute to this issue. My issue is that at random times of the day (not network bandwidth or load driven that I can tell from xymon) the Apache CPU maxes out the host (4 cores @ 100%) and the last good apache-status I could pull shows all items on the scoreboard in W "sending reply" state, and just a death cycle of workers being consumed and more Ws till no response at all. The only way to recover is to stop Apache and most of the time kill one of the processes that wont stop fully then restart. I have tweaked the timeouts for the overall apache and the proxypass elements for timeout, ttl, smax, disablereuse to prevent pooling, etc but something still hangs everything out to dry. I have seen this on 2.4.10 and 2.4.25 versions and again only using purely as a reverse proxy.. There are no local databases, no local php scripts, no local session management, etc that can account for this (like some other online posts suggests when not being used as a RP) Any ideas are greatly appreciated Danny |
|