Re: Plone timeout
Chuck Theobald <[email protected]> Fri, 10 Apr 2015 08:52:07 -0700
| Newsgroups | gmane.comp.web.zope.plone.user |
|---|---|
| Message-ID | <[email protected]> |
On 04/10/2015 12:57 AM, Maurits van Rees wrote: > Yuri schreef op 09-04-15 om 11:30: >> Il 09/04/2015 09:39, Maurits van Rees ha scritto: >>> dieter schreef op 09-04-15 om 07:52: >>>> Chuck Theobald <[email protected]> writes: >>>>> ... >>>>> One of my customers ran into an unresponsive situation wherein my web >>>>> site checked out for 15-20 minutes ... >>>>> Is this typical behavior? >>>> No. >>>> Apparently, something in your system is doing something strange. >>>> >>>> To analyse situations like this, I have developped a "request monitor" -- >>>> now available as "haufe.requestmonitoring" on PyPI. >>>> With it, you can (among other things) set up a monitor >>>> for long running requests logging tracebacks (you know from >>>> the "error_log" object) for requests that run surprisingly long. >>>> Bases on these tracebacks, you get valuable information what >>>> those long runnings requests are and what they are doing. >>> Also, you can send a SIGUSR1 kill signal to the Zope process to print >>> what the threads are doing: >>> http://docs.zope.org/zope2/releases/2.13/SIGNALS.html >>> >>> So: kill -SIGUSR1 your-zope-process-id >>> >>> When you use supervisor to start the zope instance, the output may end >>> up in something like var/log/supervisor-instance-stdout.log. It may >>> also take a while for it to end up there. >>> >>> >> Works only on debug, I think. Would be useful to have this also in >> production. > I have used it in production and it works just fine. Well, there is > probably a minimum version for this, which I do not know. > > The tricky thing can be finding out where the output ends up. Do the > kill signal, wait a few seconds, and check which files in var/log were > changed since then, and look in them. > > Apparently this did not work for me, sent the signal yesterday, but no new *log* files appear in my zinstance directory. haufe.requestmonitoring sounds promising, but I do not understand parts of the instructions, in particular "monitor.zcml must be activated". An analogy would be to say "start the car": I don't know what the key looks like and I would not know where to put it even if I could get into the car. -- Chuck Theobald System Administrator The Robert and Beverly Lewis Center for Neuroimaging University of Oregon P: 541-346-0343 F: 541-346-0345 ------------------------------------------------------------------------------ BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT Develop your own process in accordance with the BPMN 2 standard Learn Process modeling best practices with Bonita BPM through live exercises http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_ source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF