[PATCH] CTDB recovery lock cleanups

Martin Schwenke martin at meltin.net
Sat Feb 20 04:22:40 UTC 2016

I'm in the process of generalising the code that checks the recovery
lock when the main CTDB daemon sets the recovery mode to normal.  This
code will then be used to:

1. Test the recovery lock when setting recovery mode to normal (as now).

2. Take the recovery lock during recovery - this will be asynchronous,
   so will no longer block if the cluster filesystem is slow.

3. Split the recovery lock into a recovery lock, held only during
   recovery, and a cluster lock to be held by the master node.  This is
   part of the general task of untangling cluster management from
   database recovery.

Instead of only using a fcntl(2) lock on the cluster filesystem, a
configurable helper will be used to take any type of cluster
mutex.  Well, any type for which a helper can be written...  :-)

I'm going to post this in a few chunks to make it easier to review.

Attached are some initial clean-ups of the recovery lock code that make
writing the rest of the code easier.

Please review and push...

peace & happiness,
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ctdb.patch
Type: text/x-patch
Size: 17136 bytes
Desc: not available
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20160220/9fdc2da2/ctdb-0001.bin>

More information about the samba-technical mailing list