My original posting:
Our Sparc 10 Solaris 2.4 NFS server runs Prestoserve 2.4.2 on ODS 4.0
2-way metamirrors. After one submirror of a metamirror became faulty
and thus was taken off-line by ODS, the other submirror's superblock
got corrupted upon a reboot.
Sun suggested 1) not to use Prestoserve on metadevices or 2) to
metaclear the good submirror before a reboot. Has everybody else
seen such problems and come with better solutions?
Many thanks go to
Glenn.Satchell@uniq.com.au (Glenn Satchell - Uniq Professional Services)
dav@ipc.litronic.com (David L. Markowitz)
Glenn said he always does "a presto -d before rebooting to make sure that
the presto buffers are flushed before rebooting," and David heard a
solution that involved use of prestoserve on the underlying real devices
to have benefits of prestoserve to the metadevice.
When presto was tried on the underlying devices, ufsdump reported many of
DUMP: Warning - directory at inode `84673' is corrupted:
".." points to non-directory inode `82945' - run fsck
On page 190 of DiskSuite 4.0 Administration Guide, it is stated
When using Sun Prestoserve on a system running DiskSuite, you should
not use Prestoserve features on metamirrors or their sunmirros.
Therefore, I turned off presto for now.
Anchi
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:35 CDT