Word on the street is that load's going to be extra high today thanks
to some print exposure in Germany. I've taken the occasion to
stress-test Coronelli running its cache server under 2.6... Browne
(which was carrying all wikis but English) was starting to have its
periodic freezes, and at least one squid crash. I switched the .235
address over to Coronelli, so it's now handling _all_ incoming traffic.
So far it seems to be handling it smoothly; I'm not seeing the freezes
that browne had. We can switch one or both addresses back to browne if
needed.
[brion@coronelli brion]$ uptime
09:57:36 up 2 days, 7:25, 8 users, load average: 1.66, 1.78, 1.23
Some vmstat fragments:
procs -----------memory---------- ---swap-- -----io---- --system--
----cpu----
r b swpd free buff cache si so bi bo in cs us
sy id wa
2 1 9996 8128 151056 884652 21 0 107 591 1358 294 13
42 35 10
1 0 9996 9216 150652 884684 11 0 153 49 1552 408 12
44 39 5
2 1 9996 8796 150748 885064 0 0 91 615 1016 214 7
42 39 12
1 1 9996 8672 150784 885164 0 0 80 236 981 217 6
37 52 5
1 6 9996 8416 150856 885364 0 0 69 784 835 249 5
34 12 49
9 2 9996 8856 150916 885508 0 0 45 824 773 251 6
27 25 43
0 0 9996 8368 150968 885796 0 0 87 303 871 235 8
31 32 29
1 0 9996 7664 151048 886464 0 0 99 368 1121 293 13
40 40 7
6 1 9996 9012 150896 885256 0 0 61 701 1358 303 14
43 36 7
1 0 9996 8356 150980 886056 0 0 128 1 1590 458 15
41 38 5
-- brion vibber (brion @
pobox.com)