[squid-users] Revisting... High TCP_MISS rate

Hi all,
  We are running Squid w/NTLM auth between our network and a 3M DSL.
The hit rate is terrible (apparently). Here is the output from
cachmgr: client_list:
Cache Clients:
Address: 192.168.0.247
Name: 192.168.0.247
Currently established connections: 1
    ICP Requests 0
    HTTP Requests 130
        TCP_MISS 126 97%
        TCP_DENIED 4 3%

Address: 127.0.0.1
Name: 127.0.0.1
Currently established connections: 1
    ICP Requests 0
    HTTP Requests 3140666
        TCP_HIT 19500 1%
        TCP_MISS 367348 12%
        TCP_REFRESH_HIT 8331 0%
        TCP_REFRESH_MISS 3750 0%
        TCP_CLIENT_REFRESH_M 8124 0%
        TCP_IMS_HIT 33529 1%
        TCP_NEGATIVE_HIT 1832 0%
        TCP_MEM_HIT 19839 1%
        TCP_DENIED 2678413 85%

TOTALS
ICP : 0 Queries, 0 Hits ( 0%)
HTTP: 3140796 Requests, 83031 Hits ( 3%)

  I have been through the list archives and tried a plethora of
suggestions. The box Squid is on has 384Mb Ram and 40Gb of HDD space.
Per several posts on the list, I have set my cache_dir thus: cache_dir
ufs /var/spool/squid 20480 65 256. I have tried clearing the client
browsers caches, etc. But the miss rate is still high. The excessive
TCP_DENIED rate seems to be due to M$ implimentation of NTLM (per this
list).
  Squid has been up for 30 days now. I would have expected the hit
rate to be much better.
  Any help/suggestions will be greatly appriciated.

Thanks,
Chris

Chris Nighswonger
Network & Systems Director
Foundations Bible College & Seminary
www.foundations.edu
Received on Thu Jan 11 2007 - 20:20:33 MST