Greetings!
On Fri, 21 Feb 1997, Ian Cooper wrote:
> Normal response time through a monitoring script we have is around 0.2
> seconds. However, it appears that the whole service slows down
> (response times typically >50 seconds) when there are ftp requests
> being made.
>
> Has anyone else seen this? (And if so, do you have a fix?)
Well, the same hapens with our proxy, and this only started after
upgrading from squid 1.1.0 to 1.1.2. The current solution is to use a
script testing the proxy every 5 minutes and if the response time is more
than 10 seconds it reinitializes squid (the average reponse time to the
proxy is more or less the same as yours, i.e. 0.2 seconds). This is not
much of a solution, but its better than not having a good reponse time
(you cant imagine the quantity of people that complains if the proxy takes
more than 10 seconds to reply).
Machine setup:
squid 1.1.5
Linux 2.0.18 with 64MBs RAM
cache_mem 48
cache_swap 1536
ftpget_options -R -A -a -t 120 -c 3:30 -l 3:30 -r 3:30 -b 3000
dns_children 10
redirect_children 5
handling an average of 6000 requests/hour
(its not a dns problem neither a redirect problem since they never get
too busy)
Reducing the available memory doesnt help since this is a problem with
ftp (which by the way takes more time to connect to an ftp server than
without using the proxy, and it hapened even when we were running ftpget
without any options)..
Any ideas?
-- ____ \ Esoterica - Novas Tecnologias de Informacao, SA :-) Goncalo Valverde /___, grumbler@esoterica.ptReceived on Fri Feb 21 1997 - 03:37:20 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:34:30 MST