[Zope-DB] [Re: Zope-DB] odd slowness in ZEOapp server

Luis Gregorio Muniz Rodriguez kreator@dei.inf.uc3m.es
Wed, 2 Oct 2002 11:41:08 +0200


> since about a week ago we started experiencing excessive delays
> in accessing the Zope Management Interface. [...]
> /home/zope/var/log/zc# ./analyze.py Mdebug.log  5
> request 145319924 took 189.0 seconds for GET /manage_menu
> request 144738612 took 10.0 seconds for GET /manage_menu
> request 144716924 took 191.0 seconds for GET /manage_menu
> [...]
That time patterns uses to be related with reverse-DNS, rather
than with Zope itself.
Watch the network traffic (using `snoop', `ethereal', `tcpdump' ...) 
that gets in and out from the ZMI: maybe, when you make requests 
from one machine, the ZMI server needs to translate the IP address 
to a name, and thus builds up a reverse-DNS request.
If no answer is collected within some time-frame (~190 seconds, in
this case?) then the request goes on as always: hence the delays.


> the ZEO_DB server lives in a private network without access to
> a dns server, but this should not be an issue as there are hosts
> entries for the machines it needs to talk to [...]
Try to double-check it, maybe it needs some addresses that 
you aren't aware of.

HTH,
--
	- kreator