TCP_HIT in Squid 1.1.X is the same as in Squid2.0??

Hello all,

 After migrating from Squid 1.1.21 on a Sun Ultra 2 Solaris 2.6 to
Squid 2.0 on the same platform and experiencing an amazing speed up of
the cache, I do not understand why the output of calamaris.pl (v 1.20)
shows TCP_HIT latencies quite higher than before, even though the user
perception does not correspond to that.

 The output from yesterday:

# TCP-Request State Request % kByte % sec KB/sec
------------------------- ------- ------ -------- ------ ---- -------
HIT 259633 24.35 1121817 13.94 13.9 0.31
 TCP_REFRESH_HIT 105098 9.86 315319 3.92 15.7 0.19
 TCP_HIT 77377 7.26 701878 8.72 24.8 0.37
 TCP_IMS_HIT 66120 6.20 77882 0.97 0.74 1.59
 TCP_MEM_HIT 9591 0.90 24974 0.31 0.02 126.99
 TCP_NEGATIVE_HIT 1444 0.14 1762 0.02 0.01 67.39
 TCP_REF_FAIL_HIT 3 0.00 1 0.00 328. 0.00
MISS 804127 75.42 6917954 85.95 40.1 0.21
 TCP_MISS 795279 74.59 6864775 85.29 39.9 0.22
 TCP_REFRESH_MISS 8388 0.79 51780 0.64 66.8 0.09
 TCP_SWAPFAIL_MISS 460 0.04 1398 0.02 35.4 0.09
ERROR 2464 0.23 9173 0.11 0.45 8.11
 TCP_DENIED 2287 0.21 2468 0.03 0.01 60.70
 TCP_MISS 177 0.02 6704 0.08 6.15 6.15
------------------------- ------- ------ -------- ------ ---- -------
Sum 1066224 8048945 33.7 0.22

 And the extrange point is with TCP_HIT that used to be around 2
seconds with Squid 1.1.X (don't worry about MISS latencies 40 secs,
that happends when your line to USA is quite overloaded)

 So my question is if the tag TCP_HIT correspond to the same cases for
Squid 1.1 and 2.0.

Thanks for any help.

Best Regards,

Javier Puche.
Received on Wed Oct 07 1998 - 04:27:20 MDT