Re: [tlug] Slooooooow down: logs, smartctl, DNS?

[Romeo Theriault <romeo.theriault@example.com>]

On Sat, Jan 15, 2011 at 13:04, Dave M G <dave@example.com> wrote:
> Romeo, Jep,
> It seems a little similar to the lack of memory problem I was
> having on my laptop, which I eventually solved by buying new RAM. I
> wonder if maybe some threshold has been crossed where the machine is no
> longer keeping up with modern software requirements...? It is an older
> machine.
>
> Here is the output from top. I hadn't really looked at this part of the
> output, because, you know, you get habitual about looking at the parts
> that have been relevant before:
>
> top - 12:44:48 up 12:04,  2 users,  load average: 1.85, 1.67, 1.57
> Tasks: 155 total,   1 running, 154 sleeping,   0 stopped,   0 zombie
> Cpu(s): 11.0%us,  1.3%sy,  0.0%ni, 87.3%id,  0.0%wa,  0.0%hi,  0.3%si,
> 0.0%st
> Mem:   1025632k total,   971528k used,    54104k free,    22120k buffers
> Swap:  6144824k total,     4332k used,  6140492k free,   578664k cached
>
> Couple things stand out to me. One is that is says there are 2 users.
> That seems odd. Also, "id" is at 87% percent, and goes as high as 95%.
> "wa", though, is only 0%.
>
> $ uptime
>  12:49:31 up 12:08,  2 users,  load average: 1.78, 1.70, 1.61

Your LA (load average) seems quite high for a box that doesn't seem to
have much going on. While a high LA can be indicative of an issue, it
doesn't necessarily make it much easier to track it down.

>
> $ free -m
>             total       used       free     shared    buffers     cached
> Mem:          1001        921         79          0         23        532
> -/+ buffers/cache:        365        636
> Swap:         6000          4       5996


You're not running out of memory. Out of your 1GB, 636MB's of it are
either, free/buffers or is cached and you're only using 4MB's of swap.
Essentially this means your computer has plenty of memory it could
free up to use if it needed it. So this isn't an issue of running out
of memory.


>
> $ vmstat 5
> procs -----------memory---------- ---swap-- -----io---- -system--
> ----cpu----
>  r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy
> id wa
>  0  0   4320  78484  24608 548484    0    0    79    14  267  416  6  2
> 92  1
>  1  0   4320  78368  24608 548608    0    0    26     2  516  853 10  0
> 90  0
>  0  0   4320  78120  24616 548840    0    0    51    14  485  816  5  1
> 94  0
>  0  0   4320  77808  24624 548992    0    0    26    12  724  942  8  1
> 90  0
>  0  0   4320  77624  24632 549216    0    0    51    10 1243 1597 12  4
> 84  0
>  1  0   4320  76980  24632 549376    0    0    26     0  900 1383 15  5
> 79  0
>  0  0   4320  76836  24660 549704    0    0    67    23 1042 1370 14  3
> 82  0
>  0  0   4320  76028  24660 550304    0    0   123     0 1176 1568 11  5
> 83  1
>  1  0   4320  76136  24668 550576    0    0    51    10 1107 1427 13  3
> 84  0

You seldom have a "running" process and never in this sample have a
"blocking" process so your cpu is practically sleeping. Your systems
not swapping, and has plenty of available memory. You're not doing a
lot of IO. Your system interrupts and context switches look to be in a
normal range and your cpu doesn't look to be too busy either.

Hmm, not looking like there's a whole lot wrong with the system from
this stand point. I'm starting to think that jep200404 might be on to
something with the dns, particularly if your using network filesystems
like nfs. If you can log onto the machine locally (i.e. via a attached
keyboard and screen) and disconnect any other machines that may be
connected to it via the network, then bring down it's network
interface, either by pulling the ethernet cable or doing a 'sudo
ifdown <interface_name>' it would be a good test to see if the
slowness goes away or not.

Also, don't know if you did yet or not but check /var/log/messages,
/var/log/daemon.log and /var/log/syslog for clues.


-- 
Romeo Theriault