Hello,

503 errors occurred when the access concentrated.
I am investigating the cause.

Pound is not used as a load balancer, and there
is only one Web server.

A large amount of no backend is being recorded
in the log of Pound by another, and the peak is almost
the same as the peak of 20* of Apache.

Is the cause interference between pound and Apache
 though it seems that 503 errors were displayed
because of no backend?
A corresponding access to no backend is not recorded
in log (MaxClients?) in the Web server. Therefore,
I think that it is not a problem on the Web server side.

Moreover, the log of Pound is recorded from the log
of Apache after a delay of one hour.
There is a delay of about one hour in the peak in
20* counters between the log of pound and the log
of Apache accurately.
Does anyone understand what has happened?

Regards,
Akiko Irie


pound.cfg
---------------------------
ListenHTTPS prd-rp.xxxx.xx.jp,443 /usr/local/etc/xxx.pem
Loglevel 4
UrlGroup ".*"
BackEnd prd-www-vip.xxxx.xx.jp,80,1
EndGroup

ListenHTTP prd-rp.xxxx.xx.jp,80
Loglevel 4
UrlGroup ".*"
BackEnd prd-www-vip.xxxx.xx.jp,80,1
EndGroup
---------------------------

> Hello,[...]

If I understand correctly you get wrong timestamps in pounds log? If so,
maybe your pound is running chrooted? If so, you may want to make
/etc/localtime available in your chroot environment.

Regards,
Simon
[...]

Thank you for your advice!!
But timestamps are correct because there is no
delay between logs at a usual access....

Are the delay of the log and 503 errors related?

What should I examine besides the log to get the cause?

Should I doubt the network?
Should I still doubt the Web server(Apache)?

My best regards,
Akiko Irie

Simon Matter ????????:[...][...][...][...]