philipp.reisner at linbit
Oct 17, 2005, 9:10 AM
Post #1 of 1
* Fixed a SMP race condition introduced in 0.7.12, that could cause
a lockup on a busy node in Primary/SyncSource state.
* In case a disc failes during a read requests, the read requests
is reissued to the peer. Now it also tells the peer that the disk
Without that fix this bug could cause a resync in the wrong direction
after a cluster restart.
* Some improvements to the documentation.
Well, it is a shame, that there was still a problem introduced with
that 0.7.12 release.
Users of SMP machines should really really avoid 0.7.13. Now everything
works, but in case you have an outage of one node, it might cause the
primary to freeze during the resync process.
: Dipl-Ing Philipp Reisner Tel +43-1-8178292-50 :
: LINBIT Information Technologies GmbH Fax +43-1-8178292-82 :
: Schönbrunnerstr 244, 1120 Vienna, Austria http://www.linbit.com :
drbd-announce mailing list
drbd-announce [at] lists