This one is a result of some testing with various pre-rc9 code. It
doesn't remove any functionality introduced in rc7 or rc8, but perhaps
better commented code. I added inactive code to experiment with
distributed read in SSI that might make sense due to the nature of hard
This one fixes:
- Possible initnode freeze if failover node crashes before coming up.
- Possible failover node signal loop preventing CLMS callback during
Patch against drbd-0.7.10 sources.
Also if you are looking into this root failover problem that stops at fsck
that I'm investigating, try the following script in place of
/sbin/ckroot.ssi on Fedora Core. See if root failover stops before "Done
fsck exiting with error code %d". In any case if you see fsck, chances
are it means DRBD finished doing its part. Though I have a feeling that
nobody here will be able to reproduce this problem. So I'm in the process
of compiling the SSI-1.2.1 kernel to test drbd against.
if [ $REQUEST = "failover" ]
echo Doing root failover
/sbin/fsck -y -t $ROOTFS $ROOTDEV
echo Done fsck exiting with error code $result