Hello,
I try to use a delay_pool with an acl using maxconn.
My goal is to limit the bandwidth of the users who launch to many
simultaneous downloads.
Here is the acl and delay_pools setup I use :
----------------------------------
# 10 simultaneous connexions : abuse
acl abuse maxconn 10
# 1 pool for abusers
delay_pools 1
# Define the class of the delay pool
# level 2 : limit on group and IP address
delay_class 1 2
# Define the pool members
delay_access 1 allow abuse
delay_access 1 deny all
# Fix limits : 1000 bytes/s for the whole group, 500 bytes for each
abuser
# 64000 bytes may be downloaded before the bandwidth is limited
delay_parameters 1 1000/64000 500/64000
----------------------------------
This setup works halfway :
- if some IP launches more than 10 connexions, it shows up in the
delay_pool and the limit applies.
- but when the connexions ends, the IP remains in the delay_pool and the
limit remains.
I'm pretty confident in the second affirmation :
- "netstat | grep | wc" tells me the connexions are closed for the OS.
- the IP is in the delay_pool several hours after the above indicates
proper connexion ending.
The only way of removing IPs from the delay_pool seems to be reloading
the configuration (resetting acl "abuse" and delay_pools...).
Configuration :
RedHat 6.2 with all updates from updates.redhat.com applied,
squid-2.3-STABLE4 with all patches from www.squid-cache.org applied.
Is my goal achievable ?
Lionel Bouton.
-- To unsubscribe, see http://www.squid-cache.org/mailing-lists.htmlReceived on Tue Oct 31 2000 - 04:53:11 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:56:03 MST