Problem with datavg hdisk

[JSiva]

Dear All,

I have a Cluster setup of BULL EPC 810 (Equals IBM M80)systems.

In which all the systems have 2 hard disks for rootvg and datavg. Yesterday due to some problem the system was not behaving properly and I was not able to predict the error. And I was not able to do smit, touch, vi, pg or any commands.

For smit it shows /var/tmp permission denied...
For touch, vi it shows permission denied...
For pg it doesn't show anything in the file fed.

So I restarted the system and it reboots properly, but after that when I check "df" the file system which belongs to the second hard disk VG datavg is not mounted.

Even in lspv command the hdisk1 was not listed.

When I manually tried mounting it, I got the message that "Volume group must be varied on; use varyonvg command."

When I tried varonvg the datavg in hdisk1 I got the following error:

0516-013 varyonvg: The volume group cannot be varied on because
there are no good copies of the descriptor area. PV Status: hdisk1 0053938ab4b938ca PVNOTFND
----------

When I ran smit diagnostics I got the following message

A PROBLEM WAS DETECTED ON Tue Aug 20 09:58:23 CUT 2002 801014

The Service Request Number(s)/Probable Cause(s)
(causes are listed in descending order of probability):

64D-129: Error log analysis indicates a SCSI bus problem.
n/a FRU: n/a 40-61-00-1,0
SCSI bus problem: cables, terminators or other SCSI
devices
hdisk1 FRU: 07K7406 40-61-00-1,0
16 Bit LVD SCSI Disk Drive (18200 MB)
pci2 FRU: 04N6907 U0.1-P1.1
PCI Bus
n/a FRU: n/a 40-61-00-1,0
Software

Use Enter to continue.

It is a hot swappable hard disk and if there is any problem in the disk there used to be a YELLOW LED glowing in the disk and now there is no LED now in the disk.

Any idea what does this mean, and changing the bay of hdisk1 will help ?

Thanks in advance

JSiva
jsnair@lycos.com

 

[mankind]

Hello,

It seems like you've a problem with your scsi device.
The varyonvg command didn't work because the system can't access the datavg disk. As you've only one disk in datavg the vg won't activate if your physical volume can't be accessed.
It might be helpful to refer to the SRN number 64D-129 provided by the diags in the maintenance guide of the scsi bay or disk as it gives a description of what the problem may be.
At this point you should know if it's a disk or bay problem.

 

[DSMARWAY]

Hi,

It sounds as if the system thought it couldn't see your disks for a while because of a SCSI bus reset , therefore your volume group went pear shaped and now is saying that it can't find the ID of your disks .

Firstly , check if the hardware is functioning properly
i see you have done a diag , and it reprots SCSI bus problems ( log a call with IBM to check this out)

2. Check the firmware levels of your SCSI card and disks
see what level they are at and whether they can be updated . To update them you'll have to stop activity on the devices connected to the SCSI card , ( do this booting of diag disk )

3. If all firmware, disks , scsi cards are o.k. you'll have to recover / rebuild your volume group . Have you got a backup of datavg ? Have a look at the following :-

http://nscp.upenn.edu/aix4.3html/aixprob/msgdnref/err0516.htm

HTH

 

[Yegolev]

You have lost a DASD. This is a good lesson to mirror disks. You should replace the disk and restore the vg from a backup. There may be ways to fix it but it is really better to replace the disk since it will not get any better over time.

If you turn of quorum you should have fewer problems with losing disks from a vg, however since you just had one disk in the vg that would not help you. The above advice is great.

I would suggest adding two more disks and mirroring your vgs. IBM Certified -- AIX 4.3 Obfuscation

 

[JSiva]

Hi all,

Thanks for all your input and it is the problem with the Hard disk, that I was fighting with.

Thanks again.

JSiva jsnair@lycos.com