[Catalyst] Resolving performance issues.

[Wade.Stuart at fallon.com]

catalyst-bounces at lists.rawmode.org wrote on 04/25/2006 04:26:43 PM:

> On Tue, Apr 25, 2006 at 04:02:24PM -0500, Wade.Stuart at fallon.com wrote:
> > > % time     seconds  usecs/call     calls    errors syscall
> > > ------ ----------- ----------- --------- --------- ----------------
> > >  31.35    0.009970        9970         1           accept
> > >  21.89    0.006963        3482         2           send
> >
> > Those look real slow to me.
>
> I'm not clear if the database (which is the target of those sends)
> effects those times.  The database is connected via a unix domain
> socket.
>
>
> > I have never used Deamhost before,  are they
> > virtualized servers or all running in the same userland.
>
> No, it's a stand alone server.  It' "managed" in that they have their
> "control panel" code used to manage the server.
>
> $ uptime
>  14:19:22 up 35 days,  3:46,  4 users,  load average: 0.25, 0.30, 0.27
>
>
> > Can you show me the output of:
> >
> > vmstat -s
>
> $ vmstat -s
>        513300  total memory
>        503568  used memory
>        218852  active memory
>        251604  inactive memory
>          9732  free memory
>         28904  buffer memory
>        242440  swap cache
>       1052248  total swap
>        175400  used swap
>        876848  free swap
>      38575201 non-nice user cpu ticks
>       1777435 nice user cpu ticks
>      12773759 system cpu ticks
>     250598511 idle cpu ticks
>             0 IO-wait cpu ticks
>             0 IRQ cpu ticks
>             0 softirq cpu ticks
>     106342906 pages paged in
>     113293833 pages paged out
>        448685 pages swapped in
>        616766 pages swapped out
>     702841386 interrupts
>    2049418928 CPU context switches
>    1142962367 boot time
>       3485309 forks
>
>
> >
> > vmstat -a 1 10
>
> $ vmstat -a 1 10
> procs -----------memory---------- ---swap-- -----io---- --system--
----cpu----
>  r  b   swpd   free  inact active   si   so    bi    bo   in    cs us sy
id wa
>  0  0 175400  13232 249860 217044    1    1     7     9    5    10 13  4
83  0
>  0  0 175400  13236 249860 217044    0    0     0     0  217   683  1  3
96  0
>  0  0 175400  13496 249604 217044    0    0     0   208  366   703  1 10
89  0
>  0  0 175400  13504 249604 217044    0    0     0     0  268   678  0  4
96  0
>  0  0 175400  13376 249732 217044    0    0   128     0  227   681  2  3
95  0
>  0  0 175400  13376 249732 217044    0    0     0     0  273   681  2  7
91  0
>  0  0 175400  13376 249732 217044    0    0     0     0  230   688  0  5
95  0
>  0  0 175400  13340 249768 217044    0    0     0    68  300   673 10  5
85  0
>  0  0 175400  13336 249772 217044    0    0     0     0  245   713  9  6
85  0
>

Before Kernel 2.5.41 IO wait was listed in idle :/  so its hard to tell if
maybe IO is blocking..




>
> >
> > nstat (think you must be root for this one.. if not go for it)
>
> I've got root.
>
> There's no nstat on this machine.  There's the sysstat package (sar,
> iostat, mpstat) though.
>
>
> $ mpstat 2 5
> Linux 2.4.28-grsec+w+fhs6b+gr0501+++p4+c4+gr6b-v6.188 (ifpserver)
04/25/06
>
> 14:20:47     CPU   %user   %nice %system %iowait    %irq   %soft   %
> idle    intr/s
> 14:20:49     all    0.00    0.00    0.00    0.00    0.00    0.00
> 100.00    186.50
> 14:20:51     all    0.00    0.00    1.00    0.00    0.00    0.00
> 99.00    207.50
> 14:20:53     all    2.00    0.00    1.00    0.00    0.00    0.00
> 97.00    233.50
> 14:20:55     all    4.00    0.00    0.50    0.00    0.00    0.00
> 95.50    195.50
> 14:20:57     all    0.50    0.00    0.50    0.00    0.00    0.00
> 99.00    215.00
> Average:     all    1.30    0.00    0.60    0.00    0.00    0.00
> 98.10    207.60
>
> Doesn't look that bad looking at these numbers.

Yeah it looks ok, althout see my note above -- IO blocking could be killing
you -- but its listed in the idle in that kernel..

>
>
> --
> Bill Moseley
> moseley at hank.org
>
>
> _______________________________________________
> Catalyst mailing list
> Catalyst at lists.rawmode.org
> http://lists.rawmode.org/mailman/listinfo/catalyst