SUMMARY: DiskSuite - Huge Logging Device Cause Hangs?

From: Gallagher, Gerald (RTIS) (gerald.gallagher@reedtech.com)
Date: Thu Mar 30 2000 - 13:35:50 CST


Thanks to:

Lynn Schaper
Seth Rothenberg
Nickolai Zeldovich

No one was able to confirm that a large logging device contributes to
DiskSuite hangs. Basically, I need to go ahead and make a new, smaller
logging device and see if the DiskSuite device still hangs. Lynn Schaper
suggested running a script that runs mpstat, vmstat, and iostat
continuously, and then examining the output after a hang.

For anyone interested, or anyone who might have more to contribute, here are
some details on the hang condition we sometimes experience:

The RAID 5 volume hangs in that it acts as if it is read-only. We are
sharing this volume through NFS, so soon the system no longer responds to
the other hosts (though it is still ping-able, of course). If we let it go
long enough, all the OpenWindows shells will stop reacting to any commands
typed into them, and we can't open any new windows. At this point, the
system often will only respond to a Stop-A. During the time that I still am
able to execute commands, metastat happily reports that everything is fine.
I can hang any shell by trying to touch a file on the RAID 5 device.
Touching a file on any other filesystem is successful.

                Sent: Wednesday, March 29, 2000 3:56 PM
                To: gerald.gallagher
                Subject: Re: DiskSuite - Huge Logging Device Cause
Hangs?

                We're not using devices this large ... but I would agree
with you and
                suggest trashing the logging device and using one at 64Mb.

                Sent: Wednesday, March 29, 2000 9:30 PM
                To: Gallagher, Gerald (RTIS)
                Subject: Re: DiskSuite - Huge Logging Device Cause
Hangs?

                I'm not sure about whether the huge logging device is
actually causing the
                hang in your case, but it should be pretty simple to check;
just trim down
                the partition to 64 megs and see how it runs? Don't even
need to move
                other partitions :)

                Sent: Wednesday, March 29, 2000 4:41 PM
                To: gerald.gallagher
                Subject: Re: DiskSuite - Huge Logging Device Cause
Hangs?

                Gerald,
                We are using "normal" sized logging devices....

                I can't give you any comment on oversize logging devices...
                but I can tell you that we have done a number of transitions
                under DiskSuite while the system was online - eg,
                changing striping, etc. If you must fix the problem
immediately,
                I believe you can metadetach the logging device and
metattach a new one.

                -----Original Message-----

                Sent: Wednesday, March 29, 2000 2:29 PM
                To: 'sun-managers@sunmanagers.ececs.uc.edu'
                Subject: DiskSuite - Huge Logging Device Cause Hangs?

                I've struck-out in the DiskSuite list and the newsgroups,
and this is starting to become critical, so perhaps someone here can offer
some advice on this DiskSuite issue...

                My predecessor created a striped mirror metadevice with a
capacity of 1.7 GB for use as trans logging device. The master device is a
200 GB RAID 5.
                The DiskSuite User Guide says:
                                "Plan on using one megabyte of log space as
a minimum, and an additional one megabyte of log space per 100 megabytes of
file system data, up to a maximum log size of 64 Mbytes. Logs greater than
64 Mbytes waste space."
                So according to the User Guide, my logging device is way
bloated. My DiskSuite is still hanging every now and then, despite the fact
that I've installed all the patches. I'm wondering if this oversized logging
device is causing the hangs, as per SunSolve bug ID 4113855.
                Anyone else using massive logging devices? Anyone think it
is a good/bad idea to trash this logging device and create a new 64MB one?
Thanks...

                Gerald Gallagher
                Systems Analyst
                Reed Technology & Information Services



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:14:05 CDT