[DRBD-user] fence-peer

Kaloyan Kovachev kkovachev at varna.net
Tue Oct 4 13:33:01 CEST 2011


As there were no answers on the list i have digged into the source code
...

>>>  2. If any code in addition to 7 is allowed - which codes will lead to
>>> unfreezing the IO and which to keep blocking it, because in case of
>>> Inquorate cluster status or fence failure it is preferable to keep it
>>> blocked. Will returning 6 in this case lead to calling some of the
>>> pri-lost handlers i.e. commit suicide?

drbd/drbd_nl.c in function conn_try_outdate_peer - all codes are accepted
and 7 is treated like 4 (now) or 5 (maybe in future versions):

/* THINK: do we need to handle this
 * like case 4, or more like case 5? */

on my opinion it should be like 5, because if the local status is not
up_to_date no IO should be allowed. Then in the case of dual primary,
blocking the IO will give a chance for the local disk to have no changes,
which may help resolving the after-sb-0pri with discard-zero-changes later

there is no way to commit suicide and exit code 6 will outdate the local
resource if disconnected or will be ignored if the connection was restored
meanwhile

returning 1 or any other code <3 or >7 will keep blocking the IO on that
resource

attached is the modified version of the outdate-peer.sh script for use
with RHCM.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: rhcm-fence-peer.sh
Type: application/x-shellscript
Size: 3878 bytes
Desc: not available
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20111004/bc7a5f06/attachment.bin>


More information about the drbd-user mailing list