Project News: OCFS2

[ Project Home  |  News  |  Downloads  |  Docs  |  Mailing Lists  |  Source Control  |  Issues ]

 

2007.11.07: OCFS2 1.2.7-1

 

OCFS2 1.2.7-1 for RHEL4 and RHEL5 is released.

 

This release includes three frequently requested updates.

 

The first is fix is for bug#5852483. This concerns a possible memory leak in the size-64 slab during asynchronous io. This bug typically rears its head during the RMAN backup.

 

The second is a new feature, local mounts. This will allow users to mount OCFS2 volumes like any other local file system, i.e., without having to startup the cluster. Refer the man pages for tunefs.ocfs2 and mkfs.ocfs2 for learning how to toggle volumes between cluster and local mounts.

 

The third change concerns the new default cluster timeouts. These new timeouts are the same as ones we have been testing with for a long time and were the ones that were the default for the first EL5 release of OCFS2 (1.2.6-1). These are:
        O2CB_HEARTBEAT_THRESHOLD = 31
        O2CB_IDLE_TIMEOUT_MS = 30000
        O2CB_KEEPALIVE_DELAY_MS = 2000
        O2CB_RECONNECT_DELAY_MS = 2000
Refer to the Cluster Timeouts section in the FAQ for more information.

 

Relevant changes post 1.2.5-6 and 1.2.6-6 are listed below.

 

DETAIL
r3063 fs   Spec file strips debug symbols from the modules
r3062 dlm  Fix double increment of migrated lockres' owner count (bugzilla#875)
r3059 fs   Free fcbs in ocfs2_cluster_lock (bug#5852483)
r3057 fs   Check for cluster locking in ocfs2_readpage (bugzilla#907)
r3056 fs   Backport Mount Local feature to 1.2 branch
r3055 fs   Detect and use MS_LOOP_NO_AOPS
r3029 fs   Adds preferred slot# mount option
r3028 fs   Wrap access of directory allocations with ip_alloc_sem (bugzilla#859)
r3014 o2cb Update default cluster timeouts
r3004 dlm  Call cond_resched_lock() once per hash bucket scan (bugzilla#862)
r3002 fs   Remove the printing of harmless ERRORS like ECONNRESET, EPIPE. (bugzilla#865)