System is so slow

[Marco Hernandez Padill]

Tis morning our HP 900/800s went very slow. The only thing I noticed in syslog is:

Nov 22 08:28:50 ppmpdmu2 LVM[6538]: /sbin/lvlnboot -v vg00

Nov 22 08:31:53 ppmpdmu2 LVM[6717]: /sbin/lvlnboot -v vg00

Nov 22 08:31:53 ppmpdmu2 LVM[6720]: /sbin/lvlnboot -v vg00

Nov 22 08:31:54 ppmpdmu2 LVM[6726]: lvlnboot -v

Nov 22 08:31:55 ppmpdmu2 LVM[6737]: /sbin/lvlnboot -v vg00

Nov 22 08:31:55 ppmpdmu2 LVM[6740]: /sbin/lvlnboot -v vg00

What does it mean??
Thank's
MPADILA

[Ajitkumar Rane]

The lvlnboot command is used to define the boot,root,swap and dump partitions using various options.The command with -v option is used to display/check these partition information.

[Madhu Sudhan_1]

The data given is insufficient to find out why the system is slow. Just the lvlnboot -v doesn't cause the system to go slow. May be you have to look at OLDsyslog.log incase if there was a reboot.

...Madhu

[Philip Chan_1]

Marco,

Your disks may be very busy hence slowed down the system. Check it out by the "sar" command.

eg. sar -ud 2 9

If my assumption is true, next thing you should do is to find out what triggered these disk I/O.

[John McWilliams]

Marco - do you have top and glance. these are useful tools for determining what is causing the system to be slow. glance is good as you can look at specific things eg disks, memory, cpu. Check the man page on glance as this has lots of good instructions on this

John

[Marco Hernandez Padill]

Checking the listener log (from Oracle) I found this:

TNS-00530: Protocol adapter error
HPUX Error 233: No buffer space available

And checking the kernel parameters I have:
bufpage=0
nbuf=0
dbc_max_pct=50

What's the problem??

[Albert E. Whale, CISSP]

I have two brief suggestions:

1. Check for DNS or Name Resolution working.

2. I would tune the Kernel. At a MINIMUM, I would recommend the Monolithic Database Server. Depending on your Oracle DB and Application Requirements, you may also need to increase a few parameters to meet their needs as well.

I would also like to ask, how much Memory and Swap are on the Server.

[Al Langen]

When the system is running slow: run top or, if you have it, glance.
Next run swapinfo -t
Next run netstat -rn looking for new routes by discovery to a device that is not a router or gateway.

[Bill McNAMARA_1]

A slow system can be caused by a thousand
things. What might help us is to help you
is if you send some info.
Your kernal tuning params I believe are the
defaults. However you'd only tune outside
default configurations if for example you
do more i/o than processing etc..
You really will get much more performance
out of your system by checking the hardware
connectivity of perhiferal disk devices and
'balancing' traffic along the available
paths. ie some lvm primary links on 8/0 and
some on 8/4 some on 8/8 etc.. not all on 8/4.
You may not be doing this but its a
recommendation. Also tell us if you're using
raid, send the output of the mount -p
command to make sure that you're mounting
the fsystems with okay parameters, check that
your not paging to much and that swap is okay
with swapinfo (more swap is better). Make
sure secondary swap is created with
no bad block relocation and contiguous
allocation policy. After that, or
rather before that, make sure from swlist
that you're not running on an unpatched
sw.

If you really want to be thorough with info
gathering, run the nickel script
www.grc.hp.com/docs/nickel/
(or search for nickel from the www.grc
and send the output to us in attachment.
Should be less than a hundred K zipped.

Cheers,
Bill