[pmwiki-users] PmWiki seems to "hang" on overloaded boxes
Joachim Durchholz
jo at durchholz.org
Sun May 22 07:11:22 CDT 2005
Hi all,
I'm having some trouble with a PmWiki installation that may, in fact, be
more related to Apache/Linux configuration issues, but I haven't been
able to exactly pin the problem down so my main question is whether
anybody else has seen similar behavior.
Here's what happened: PmWiki suddenly started to "hang", i.e. it would
not respond at all, or display only part of the output and "hang". With
"hang", I mean that the browser still has the data-is-forthcoming
animation, but no new data ever arrives. (For an "forever" value of one
or two minutes.)
I have done some server twiddling, reducing the number of processes that
Apache preforks (it used to prefork an unreasonable number of processes,
only to have them swapped out to disk - which is *not* what preforking
is about *gg*). Either that or a decrease in server load has returned
PmWiki to normal, sane behavior.
Here are my questions related to this issue:
1) Has anybody seen similar behavior?
2) Does anybody have a clue why PmWiki hangs, instead of returning an
error message or simply abruptly terminating the connection? Or is it
normal that the browser continues to wait for data when a PHP process
was killed due to CPU/memory over-use?
3) Does anybody know how to diagnose server load well? (I have a Debian
Linux.) I need to see what processes go to swap, how much of each
process got swapped out, CPU usage, and similar data. "top" seems to fit
the bill, but I find it difficult to interpret the data (this may just
be a case of RTFM, but I'd like to hear whether there are friendlier
alternatives before delving too deeply into the "top" manual page - it's
over 1000 lines...) "top" also seems to be limited to giving a
current-situation snapshot, no overview over time - and I'd like to do a
post-hoc analysis of how loaded the system was at, say, three o'clock in
the morning when I'm in bed :-)
Regards,
Jo
More information about the pmwiki-users
mailing list