mirror problem

[Javier Ortiz Guajardo]

I have this problem, I have a disk in vg00 and it contains the /home /usr and a swap device. This disk is mirroring in c4t5d0. But yesterday the c4t5d0 fails with errors like mechanical positioning error, software configuration error, and the message The device was not ready to process request when it received a request from the device driver because it is in the process of becoming ready.
the system hang and the ioscan display the disk claimed, but the system does not respond to login attempts and some internal commands for users that has already login.
The question is, why was the system hangs if the disk mirror fails.

why the disk stay becoming ready?, i need to do something?

Thanks.

[Patrick Wallek]

The first thing I'd do is call the HP hardware folks and have them get someone out with a replacement disk.

I'd also check your mirroring setup and make sure that it is properly setup for those LVOLs. It sounds like your mirroring is not doing what it is supposed to be doing.

[A. Clay Stephenson]

This is just a guess. Are both your disks on the same bus? If so, bus resets will then effect both mirrors. You should be mirrored not simply on separate disks but on separate controllers as well.
It is also possible that evenn if you are mirrored on separate buses that you have problems with more than one disk drive. Also, are all lvols (including swap) in vg00 mirrored? Finally, were users trying to access non-vg00 data/applications that were not on mirrored LVOL's?

[Javier Ortiz Guajardo]

i??m agree the disk must be replaced, but i i want to find why the system hangs, and clay you??re right when the disk fails (c4t5d0) the system do a bus reset and send a message from c4t6d0 fails but it recovers succesfull, but not the disk c4t5d0.
the primary disk are in the address (1/0/0.5.0 c0t5d0) (1/0/0.6.0 c0t6d0) while the mirror disk are in (5/0/0.5.0 c4t5d0) and (5/0/0.5.0 c4t6d0)
in that way i suppose they are in different bus. am i right?

i hope you help me to find why the system hangs.

thanks so much.

[Rajeev Shukla]

You are right, both the disks are on different SCSI buses. Now if the mirror disk fails completly then you wont have system hanging. In your case the disk goes offline and then comes online which takes most of the system resources in detecting the disk which is making the system slow. This could be the problem with the power supply of the disk or even the connects could be the problem. The best solution is to get the HP engineer replace the disk and remirror the Disk.

Rajeev

[A. Clay Stephenson]

You seem to be mirrored correctly from a separate bus standpoint (at least vg00) but that still leaves open the question are all LVOL's mirrored and are there any other VG's with LVOL's that are not mirrored?

At this point you need to do some vgdisplay -v and lvdisplay -v command and look at the status of both volume groups and logical volume extents. Man vgdisplay and vgdisplay for details.

I hope that all of your drives are hot-plug. If so, you can replace then without ever shutting down. If your are configured correctly then your system should keep chugging along and barely even care that you have a bad drive. In many cases, the system will be more stable if you can simply remove ('on the fly' if it's a hot-plug) the flaky drive. I have enough drives that I routinely replace a few failed disks every month and have never lost data or even had to shutdown. Moreover, no significant performance impacts occur during this time. I would really start looking at the possibility that you have more than one failed drive or an unmirrored LVOL. Remember, you are not mirrored at the physical disk level but only at the logical volume level. It is even possible that you have mirrored LVOL extents on the same drive which could cause exactly the kinds of problems that you are seeing because both mirrors are on the same physical device.

[Javier Ortiz Guajardo]

thanks, i will check the lvols mirror configuration, but i have one more question.

i think the disk could not be operational because stay in the message "the device was not ready to process request when it received a request from the device driver because it is in the process of becoming ready"

do you know something about this message "becoming ready"

thans so much again.

[T G Manikandan]

Whatever happens,
the first step is to look at the system logs.

/var/adm/syslog/syslog.log and /var/adm/syslog/OLDsyslog.log files.

If there seems to be problem with the system then there should be plenty of messages from the OS into these logs.

Revert with the log messages