Bad block allocation

[Ray Herbig]

A while back someone posted about EMC disk arrays and LVM bad block allocation. The guy asked whether or not he could turn off bad block allocation and whether or not it was possible to check if something had been written to the bad block directory. The second part wasn't answered which is my question. Is it possible to tell if an entry has been made in the bad block directory, and if so is there some way to "clear" it out.

Thanks.
ray

[Vincenzo Restuccia]

Try with fsck

[Derek Nicholls]

The tool that you require is called bbdir. This should be available from HP or EMC (It was 3 years ago). You may encounter difficulty in obtaining it if you do not have a specific problem (It's been some time since I last tried!)
Bad Block Relocation on HA Storage Arrays should be turned OFF. This can be acheived by
'lvchange -r N 'lv_name'

Entries are available to explain this topic further on http://www.dutchworks.nl/htbin/hpsysadmin
which may help you further.

[Ray Herbig]

Ok. So can any harm come from turning bad block allocation off?

thanks.

[Global Unix Team]

Does anyone know whether bad block allocation needs to be turned off (set to NONE) for the FC60 or VA7100 disk arrays?
Thanks

[john kingsley]

I think you mean "Bad Block Relocation". EMC recommends that you disable this feature on lvols residing on their SAN. This can be done by:
lvchange -r n /dev/vg##/lvol#

[Prashant Zanwar_4]

I am not sure of what EMC says about it. If so they must be having some internal mechanism which will take care of this technique.
Bad block relocation is used in system to mark the bad blocks on your media and move the data from them to good area. And then it becomes a reference for system that it should not right to bad blocks.
It is recommended to turn off the bad block on root VG so as to detect disk failure whenever occur.
Whereas on data volume it shall be turned on so as not to loose data.
Please check with EMC what do they do internally with bad blocks.
Regards
Prashant

[Geoff Wild]

No harm can be done by turning BBR off - this is a recommendation from EMC. My understanding is, is a spot on the disk is going bad, that block is marked in the frame, and the frame calls home (EMC) to have disk replaced. Redundancy in the frame keeps your server happy - and doesn't even know something was wrong..

Rgds...Geoff

[David Child_1]

John,

EMC's old recommendation was to prevent bad block relocation using 'lvchange -r n', but if you check current EMC documentation and/or cases you will find that they now say that bad block relocation should be set to Disabled via 'lvchange -r N'.

The bad block is handled on the EMC frame. Unfortunately I found out the hard way that using 'lvchange -r n' was not the way to go. We did experience a bad block and I ended up with a problem on one of my logical volumes. The DBAs were reporting I/O errors on one of their databases. I confirmed this by checking syslog and then using 'dd' to read the logical volume.

Once I confirmed the I/O errors I opened a case with EMC, but they didn't find anything on the frame. I suspected a bad block issue after I found the 'lvchange -r N' information on EMCs web site, but couldn't confirm until I tried removing and readding the alternate link for the PV this LV resides on. When I tried readding the alternate link I got this error:

PV /dev/dsk/c47t2d7 has the following bad block/s with allocated alternates.
Please run pvcreate [-B] -b and specify these bad blocks before re-running
the command again. Should you wish to override this operation, re-run
the command with the -f argument. In doing so, you may encounter data
corruption.
Bad blocks: 28666504 28666568

Luckily the DBAs had a current backup as I was unable to fix this problem.

David