[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20040526: bigbird raid reconfig

>From:  Gerry Creager N5JXS <address@hidden>
>Organization:  Texas A&M University -- AATLT
>Keywords:  200405270011.i4R0BjtK010667 LDM Linux RAID JFS

Hi Gerry,

>Are you still on?

I am back on now (11:30 pm my time).

>I'd like to try to remake the RAID.  We'd lose some 
>data but I suspect it would be a faster and more effective method than 
>waiting for fsck.jfs to complete, and the raid to resync.
>
>One reason for this is I just found and cleaned up a fubar or 2 in the 
>/etc/raidtab file.  This could have affected jfs performance.
>
>Thoughts?

If you found something that will speed up the performance of the RAID,
then I say to remake it!  We have been scratching our heads trying
to figure out why the performance of the RAID on bigbird was so poor.
For instance, it was taking 6 hours to remove a full day of NEXRAD
Level III data (NNEXRAD) and over 3 hours to remove a day of CRAFT
data.  Remaking the RAID will only lose data, and that is fine by
me since we are really trying to learn how to tune a RAID under
Linux anyway.

So, I was going to logon to bigbird and mount the RAID filesystem
/data if fsck.jfs was finished.  I will now wait for you to remake
the RAID and do the mount.  As far as I can tell, you should then be
able to start the LDM and let data flow in.

>From: Gerry Creager N5JXS <address@hidden>
>Date: Wed, 26 May 2004 20:56:53 -0500

>> Interesting...  Did you run fsck (or variant) to get things patched
>> up before remounting the RAID filesystem?

>Yes.  Took a few (10 or less) min.  This time was taking a lot longer.

fsck.jfs was taking a considerable longer amount of time.  I was
assuming that this was at least in part caused by the size of your
RAID.

re: restart of LDM

>Was, or should have been, embedded in rc.local

I didn't see it there:

#!/bin/sh
#
# This script will be executed *after* all the other init scripts.
# You can put your own initialization stuff in here if you don't
# want to do the full Sys V style init stuff.

touch /var/lock/subsys/local

The setup I installed is a script named ldmd that is started at
run level 5 with priority 95 (so it comes up close to last in
the bootup sequence).  The script is the standard one we use
at the UPC on our Linux machines running the LDM.

re: the load average did go up to 400...

>Yep.

re: running fsck.jfs

>It's taking a lot longer than it ever has before....

>I'll be writing most of the night.  Proposal time.

If you are still up and looking to take a break from proposal
writing, you might try remaking the RAID.  I just logged onto
bigbird and see that fsck.jfs has finished.

I'll check back on bigbird in the morning...

Cheers,

Tom
--
+-----------------------------------------------------------------------------+
* Tom Yoksas                                             UCAR Unidata Program *
* (303) 497-8642 (last resort)                                  P.O. Box 3000 *
* address@hidden                                   Boulder, CO 80307 *
* Unidata WWW Service                             http://www.unidata.ucar.edu/*
+-----------------------------------------------------------------------------+