Actually - I found something.

 

This solaris machine is a virtual machine, and I shutdown one of the other
virtual machines.  As soon as I did that, everything was awesome.  I am not
sure if there's simply too much load on the CPU, or if the other VM just
needs vmware tools installed, but I know it's related to the fact of
competing for resources against another VM.

 

 

 

From: [email protected]
[mailto:[email protected]] On Behalf Of Edward Ned
Harvey
Sent: Saturday, August 15, 2009 8:42 AM
To: [email protected]
Subject: [osol-discuss] Performance Problems - how to kernel monitor

 

I'm having terrible performance problems right now (trying to FTP a large
file from a solaris server, and only getting 8Mbit on 10/100 LAN)

 

This is my only clue so far:

[root ~]# top -Icb

last pid:  5510;  load avg:  7.23,  7.71,  5.30;  up 0+11:25:06
08:32:03

48 processes: 45 sleeping, 1 running, 1 zombie, 1 on cpu

CPU states:  0.0% idle,  1.6% user, 98.4% kernel,  0.0% iowait,  0.0% swap

Kernel: 108 ctxsw, 4 trap, 1455 intr, 272 syscall, 4 flt

Memory: 1024M phys mem, 49M free mem, 512M total swap, 510M free swap

 

   PID USERNAME LWP PRI NICE  SIZE   RES STATE    TIME    CPU COMMAND

  1907 root       1  33    0 3392K 2248K run    441:19 78.50%
/usr/sbin/in.ftpd -a

   852 noaccess  17  59    0  172M   96M sleep   14:23  9.32%
/usr/java/bin/java -server -Xmx128m -XX:+UseParallelGC
-XX:ParallelGCThreads=4

  5510 root       1  33    0 2816K 1512K cpu      0:01  8.72% top -Icb

  3221 root       1  59    0  164M   23M sleep    3:10  0.64%
/usr/X11/bin/Xorg :0 -depth 24 -nolisten tcp -nobanner -auth
/var/dt/A:0-NvayBb

   670 root       1  59    0 3356K 1320K sleep   44:29  0.53%
/usr/lib/vmware-tools/sbin/amd64/vmware-guestd-binary --background
/var/run/vmw

  3259 root       1  59    0   19M   13M sleep    2:13  0.42% dtgreet
-display :0

     7 root      15  59    0   13M 8584K sleep    1:04  0.21%
/lib/svc/bin/svc.startd

   544 root       1  59    0 7452K 1712K sleep    0:09  0.10%
/usr/lib/sendmail -bd -q15m -C /etc/mail/local.cf

   329 root       1  59    0 1420K  656K sleep    0:01  0.03% /usr/lib/utmpd

   314 root       4  59    0 7972K 2792K sleep    0:15  0.02%
/usr/lib/inet/inetd start

   633 root       1  59    0 2652K  564K sleep    0:10  0.02%
/usr/sbin/vmware-memctld --background /var/run/vmware-memctld.pid

Notice:  in.ftpd is consuming nearly all the CPU, and the system is spending
98% in kernel time.  No iowait.

 

This is a simple IO request.  I don't think the system should be processor
bound.  I am curious why the system is spending all its time waiting for the
kernel in the processor.  I have no skills on dtrace.  Could anybody offer
me any advice on how to figure out what the system is doing . why it's
taking so long . etc?

 

The system has been in this state for the last 10 hours, and the progress
meter predicts another 10 hours, although my manual calculator says another
5 hours.

 

Thank you for any assistance.

_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to