0

I have something weird going on with my workstation, like see it says 42.1%wa, whenever I get to this it becomes so slow, any ideas what can be done, how I can identify what's causing it and all?

top - 15:44:02 up 93 days,  4:29, 16 users,  load average: 1.94, 1.86, 1.24
Tasks: 206 total,   1 running, 203 sleeping,   2 stopped,   0 zombie
Cpu(s):  3.8%us,  4.2%sy,  0.0%ni, 49.6%id, 42.1%wa,  0.2%hi,  0.2%si,  0.0%st
Mem:   3969460k total,  3585412k used,   384048k free,    65896k buffers
Swap:  2031608k total,     9200k used,  2022408k free,  1701636k cached

update: i just did import of MySQL backup (mysqldump) to both systems... load went to 3-4 on both of them, in fact slower system was running lower load average then faster system but not by much.. anyway here is output...

[root@wcmisdlin02]# time gzip -cd mysqldump-uftwfdb1.uftmasterad.org-phplist-20110223.sql.gz | mysql -p phplist2
Enter password:

real    159m34.126s
user    2m14.677s
sys     0m10.757s
[root@wcmisdlin02]#

[root@wcmisdlin03 tmp]# time gzip -cd mysqldump-uftwfdb1.uftmasterad.org-phplist-20110223.sql.gz | mysql -p phplist2
Enter password:

real    463m15.483s
user    3m53.825s
sys     0m23.454s
[root@wcmisdlin03]#

so on one it took 159m and another one it took 463m oh yeah... and here is htop from one of the machines (faster one) when it's "idle" http://pastebin.com/qG1MQCA0 and when it's doing import http://pastebin.com/qRZpQKt3

alexus
  • 13,112
  • 32
  • 117
  • 174

6 Answers6

1

Get htop, its a little easier to figure out whats going on using this tool. Also it would help to see the process table, paste output of ps aux.

ztron
  • 317
  • 1
  • 8
  • htop is installed ps aux output posted here : http://pastebin.com/qG1MQCA0 – alexus Feb 17 '11 at 21:25
  • I assume the output of ps is from the time when the machine is slow? :) – ztron Feb 17 '11 at 21:26
  • don't see anything crazy going on.. you have apache and mysql running, could other users access your machine for resources? – ztron Feb 17 '11 at 21:30
  • netstat shows no activity, so not much going on in that regards, yet top suggests otherwise ... – alexus Feb 17 '11 at 21:33
  • @ztron, unfortunately no its a bit after that.. but i'll try to grab both of them next time it happens... – alexus Feb 17 '11 at 21:34
  • Output of ps is slow when you have high disk IO because when you run command it does stat on current directory and it takes time to read from disk pegged by high IO. I suggest moving to some pseudo dir like /proc ; then ps will run fast ;-) # cd /proc; ps fuxaw – Hrvoje Špoljar Feb 17 '11 at 22:54
1

If you suspect an I/O hog process, you can use iotop to find it.

Recent versions of htop also have per process I/O stats, but you might need to configure htop to show that.

halp
  • 2,208
  • 1
  • 20
  • 13
  • i did install htop, so i'll try to take a screen shot of htop when next time it happens .. – alexus Feb 22 '11 at 21:56
  • here is htop output http://pastebin.com/qRZpQKt3 – alexus Feb 23 '11 at 16:55
  • Does not include I/O stats by default like I said. To include them: press F2, Setup->Columns, in Available Columns scroll down and find the IO_* ones. After adding them, then sorting the processes by IO, you will see which ones do it most. – halp Feb 23 '11 at 21:18
1

If you think its IO load, use lsof to determine what hooks are into your local file system.

SuperBOB
  • 460
  • 3
  • 7
0

Have a read of the man page for top: you can re-order the process list by column, this might help to identify what's hogging the system, or what type of thing is causing pain.

Your symptom(s) might be driver or ACPI related, but it's unlikely. You can check with a liveCD (of your release and if needed a different one) to see if it has the same trouble.

Also, use netstat to check your network I/O isn't going crazy. And maybe install systat to give you plenty of I/O data. Manuals will need to be read :-)

DutchUncle
  • 1,265
  • 8
  • 16
  • i'm pretty confident that this has nothing to do with networking, however I'm also pretty sure that it's somehow linked to my disk I/O, I was doing something with MySQL at the time, but I have another workstation I run MySQL there as well, it works *much* faster and specs wise it's exactly same workstation as the one I'm having issues with... – alexus Feb 17 '11 at 21:40
  • There's a recent (today Feb 17th) thread/Q on MySQL performance on serverfault, go look for it :-) – DutchUncle Feb 17 '11 at 22:04
  • do you mind posting a link? since you referring to a particular thread? – alexus Feb 17 '11 at 22:45
  • Read through the answer by cwebber: http://serverfault.com/questions/236828/dual-quad-core-xeon-cpu-but-still-high-load – DutchUncle Feb 18 '11 at 18:25
0

Are you running something that takes a lot of disk IO or memory? High % of wait could mean lots of disk access or lots of swapping. If you run NFS, maybe network congestion? To monitor disk IO, use iostat. (iostat is in the sysstat package on RH, and it might not be installed by default)

James
  • 31
  • 2
  • as you see in stats, swap is not being used and user only using 3.8% and machine is almost 50% idle where the other 50% taken by %wait, I am working with disk but nothing that out of ordinary, same task on another workstation doesn't take 50% wait, it happens within minutes while on original workstation a lot longer... – alexus Feb 22 '11 at 21:55
0

Is your CPU throttling and not properly unthrottling?

Monitor /proc/cpuinfo when it is 'slow' and see.

Doug
  • 646
  • 3
  • 8