Pillar Axiom 600 Storage System
 

PLA-Support>Sun Systems>DISK>Axiom>SN-DK: Ax600
 

Applies to:

Goal

 This document explains meaning of the "Volume Bad Blocks" error and recovery options.

Solution

 Axiom can report an alert  "Volume Bad Blocks" about a volume if it detects and bad block on the volume during a migration to different storage domain via Axiom tools, "Clone LUN", or "Copy LUN" operation.

An example of the Alert here and there is no proposed action to clear or recover this bad block: 

The source of the error should be a bad block recorded at brick (RAID) level during or before migration.  Once RAID has a bad block for the storage under a LUN, if you copy or migrate the data in a LUN using the Axiom, the Axiom must make sure that no host would be given the false impression that there is nothing wrong with the data on the LUN.   In order to do this, as you copy/migrate, the Axiom takes the RAID bad block entry and converts it to an Array Manager Bad Block entry at higher level for that exact same Logical Block Array (LBA).  Host would still see a Media Error on a read, but on a write, the Array Manager bad block entry would be cleared. Both bad block records and their details are not available for end user and can readable only by Oracle TSC. The only way to get rid of those bad blocks is to write that LBA from a host, with correct good data.

To recovery:
1.  Turn off all new snapshots for the LUN reporting bad blocks via setting the respository space to 0 at  "Data Protection" screen on GUI.

2.   Do not use any Axiom based utility to move the data from the source to target LUN/bricks.

3.  Create new LUNs with same size and QoS settings and use ONLY host based utilities to take data from the source and put it on the target LUN/bricks.

4.  Use host based data integrity utilities to check and recover the data on the LUNs on the target LUN or bricks to the best of their ability.  Reported Bad blocks may not be in the user area and host based tools may not report any error against them. This way the bad block will remain in source LUN/bricks and will not be propagated to target LUN/bricks.

5.  For any bad block on the newly created LUN/bricks where a Media Error is returned, only a host based write of data will clear that condition.  That means host based utilities did not help and only restore from a clean/good backup will clear bad blocks.

References

This solution has no attachment