LVM entry in messages

[Layne F Bouda]

Hello,
Is it normal to have this entry in your messages log:
LVM: PV 0 has been returned to vg[8].
LVM: PV 0 has been returned to vg[8].
LVM: PV 0 has been returned to vg[6].
LVM: PV 0 has been returned to vg[1].
There are 10 vg# on the system, they all appear in the log.

[harry d brown jr]

See this thread:

http://bizforums.itrc.hp.com/cm/QuestionAnswer/1,,0xe6eec6af36b7d5118ff10090279cd0f9,00.html

Did you experience a powerfailure?

live free or die
harry

[Wim Rombauts]

It should mean that there was something wrong with a physical disk (Physical Volume 0). It is OK again, and now has rejoined the LVM configuration and has been added again to all volume groups.

Does this make sense ? Are there other LVM messages in syslog.log that confirm this ?

[Layne F Bouda]

Yes there are entries in the syslog.log to confirm, They do mention a POWERFAIL
but I am sitting here watching the log grow and we are not getting any fluxuation in power.
The responce center has informed me that I need a firm ware upgrade. Could this really help.

[harry d brown jr]

An upgrade to your firmware wouldn't hurt, but I would have HP do it, because if it fails it could fry your machine. You could also be experiencing a power supply failure. Also, power fluxuations (both spikes and sags) might not be detectable without power monitoring equipment.

live free or die
harry

[James R. Ferguson]

Hi Layne:

It is indeed possible that this is *not* due to any fluctuation in power but rather very high disk I/O. That is, an I/O has not completed in the TIMEOUT period currently used for the disk.

You can increase the TIMEOUT to see if this is the problem without impacting performance:

# pvchange -t 180 /dev/dsk/cXtYdZ

You can find the current timeout value by doing a 'pvdisplay' for the disk in question. Look for the IO Timeout value.

The device name in the message you describe points to the affected disk. The two
digits immediately after 0x are the hexadecimal equivalent of the decimal
major number for the device (the driver's number). The next three digits are the the hardware path.

Regards!

...JRF...

[Roger Baptiste]

Layne,

Yes, it is "normal" , in the sense that your LV's are being switched over to the alternate PV links. I have many such messages on few of my systems. As long as there are no "lbolt" errors (which implies problems with accessing the disk), it is ok.

The reason it is happening is because there is a timeout value for each disk (default 30 sec?) and if the system cannot access it by that time, it switches on to the alternate link as defined in your VG configuration.
(VGdisplay -v will show the primary and alternate paths). That's when this messages are recorded on the syslog file. Again, when the primary link seems accessible, it will switch over back to it.

To stop this switching, it is preferable to increase the timeout value to 180 seconds.
But, pls note this should be done for diskmodels where you are getting this messages.
For instance, i have timeout of 180 set for EMC disks and XP 256. (not for internal disks and couple of other vendor disks we have on the site).

pvchange -t 180 /dev/dsk/xxx

is the command.

HTH
raj

[Sanjay_6]

Hi Layne,

This is not a normal message. It looks like there might be a disk problem with the first disk in your 8th VG. do a vgdisplay -v for that VG and replace the disk if you can. This could be a timeout probelm with the disk too.

Take a look at these threads for more info,

http://us-support2.external.hp.com/cki/bin/doc.pl/sid=6850ce3b15e1caf07c/screen=ckiDisplayDocument?docId=200000048469426

http://us-support2.external.hp.com/cki/bin/doc.pl/sid=6850ce3b15e1caf07c/screen=ckiDisplayDocument?docId=200000024602982

Hope this helps.

Regds

[Layne F Bouda]

I have reset all disks to a timeout of 180 last Thursday night.
Good call on that.
I have looked at the set_fixed -l and found a number of Items that the system claims are down. But I can see that they are up and running. How do I use set fix to return the status to an 'UP' status?
set_fixed -n ??
Could the system be reading these status and calling out erroneuos errors?