On 14 Nov 2002 at 11:03, Henrik Steffen wrote:
> vmstat 1 5:
> procs memory swap io system cpu
> r b w swpd free buff cache si so bi bo in cs us sy id
> 1 8 1 60 4964 5888 309684 0 0 176 74 16 32 25 9 66
> 0 6 3 60 4964 5932 308772 0 0 6264 256 347 347 13 9 78
> 0 5 1 60 4964 5900 309364 0 0 9312 224 380 309 11 6 83
> 1 4 1 60 5272 5940 309152 0 0 10320 116 397 429 17 6 77
> 1 4 1 60 4964 5896 309512 0 0 11020 152 451 456 14 10 76
> w:
> 12:04pm up 2 days, 17:44, 1 user, load average: 10.28, 7.22, 3.88
> USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT
> root pts/0 condor.city-map. 11:46am 0.00s 0.09s 0.01s w
> this is when things begin to go more slowly....
Two things immediately noticable.. Load average and block ins..
Either your disk write BW is saturated or CPU is too full, which I believe is
the case. HAve you ever got faster write performance than 12K blocks say? Disk
BW may be a bottleneck here.. Are they IDE disks?
Besides almost transactions are insert/update.. And if you have 11K blocks per
second to write.. I suggest you vacuum analyse most used table one in a minute
or so. Decide the best frequency by trial and error. A good start is double the
time it takes for vacuum. i.e. if vacuum analyse takes 60 sec to finish, leave
a gap of 120 sec. between two runs of vacuum.
You need to vacuum only those tables which are heavily updated. This will make
vacuum faster.
HTH
Bye
Shridhar
--
Nouvelle cuisine, n.: French for "not enough food".Continental breakfast, n.:
English for "not enough food".Tapas, n.: Spanish for "not enough food".Dim Sum,
n.: Chinese for more food than you've ever seen in your entire life.
