pvfs2-db-multi-server.patch: ---------------------------- This patch fixes a problem that we have seen a few times but only recently figured out how to reproduce. The situation is that one machine is running a pvfs2-server and then a second pvfs2-server is started on the same machine after a heartbeat failover. Sometimes the first pvfs2-server process goes into an error state with a DB_RUNRECOVERY error. It appears that the cause of this is a conflict between the shared memory regions used by Berkeley DB for the two processes. This patch makes two changes. First it updates the mechanism used to determine a unique shm key for each server so that it is based on an ordering of the available servers rather than port number. The port number is a problem nowadays because two daemons on the same machine can use the same port if the TCPBindSPecific option is used. The second change is to have trove use unlink rather than dbenv->remove to clean up old __db.* files on startup. The problem with the remove function (particularly with the FORCE flag) is that if there is an old __db.* file for the new daemon that happens to point to the same shm id as the already running daemon, it will cause corruption. The Berkeley DB documentation indicates that deleting the __db.* files is a valid environment cleanup mechanism for operating systems that allow it. In summary, the intention of this patch is to make sure that there is no collision on shm id or shm key when two servers run on the same machine and same port. The ipcs command line utility (with -p and -m options) was particularly useful in tracking down this problem.
pvfs2-heartbeat-updates.patch: ------------------------------ This patch has three updates/bug fixes to the heartbeat examples. It updates them to work correctly with heartbeat 2.1.2 (while still being backwards compatible with 2.0.8). It also updates the example PVFS2 resource script to perform a kill -9 on shutdown if the original kill fails. Finally, there is an example of how to monitor the status of a particular outlet on a power strip. -Phil
pvfs2-db-multi-server.patch
Description: Binary data
pvfs2-heartbeat-updates.patch
Description: Binary data
_______________________________________________ Pvfs2-developers mailing list [email protected] http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers
