bareos: master cf07f93d

Author Committer Branch Timestamp Parent
mvwieringen mvwieringen master 2014-02-11 09:49 master 6738338a Pending
Affected Issues  0000276: SD to SD replication makes SD crash
Changeset SD to SD replication makes SD crash

Problem is that we wait up to 30 minutes for the SD-SD replication
job to start but it could be that the total migration/copy run takes
longer then 30 minutes to finish all jobs. The current code doesn't take
this into consideration and doesn't check if there is actually a remote
SD that connected for doing the replication.

So fix is to not continue when sd replication socket is not connected.
And also remove the timeout on starting a SD-SD replication session.
The normal FD-SD connection is protected with a timeout so we don't hang
when a FD never connects. As we support canceling a storage Job from the
director and the director should cleanly cancel the storage job any way
when it fails the copy or migration job it should be no problem.

In a normal backup/restore there are 3 daemons involved e.g. director,
storage daemon and file daemon but with migration the director controls
everything and controls the at maximum two storage daemons.

Fixes 0000276: SD to SD replication makes SD crash
mod - src/stored/dir_cmd.c Diff File
mod - src/stored/sd_cmds.c Diff File