If an item was locked we should not update xa_last_pushed_lsn and thus skip it when restarting the AIL scan as we need to be able to lock and write it out as soon as possible. Otherwise heavy lock contention might starve AIL pushing too easily, especially given the larger backoff once we moved xa_last_pushed_lsn all the way to the target lsn. Signed-off-by: Christoph Hellwig <hch@xxxxxx> Reported-by: Stefan Priebe <s.priebe@xxxxxxxxxxxx> Tested-by: Stefan Priebe <s.priebe@xxxxxxxxxxxx> Index: xfs/fs/xfs/xfs_trans_ail.c =================================================================== --- xfs.orig/fs/xfs/xfs_trans_ail.c 2011-10-06 10:31:29.855243550 -0400 +++ xfs/fs/xfs/xfs_trans_ail.c 2011-10-06 10:31:55.832744112 -0400 @@ -452,7 +452,6 @@ xfs_ail_worker( case XFS_ITEM_LOCKED: XFS_STATS_INC(xs_push_ail_locked); - ailp->xa_last_pushed_lsn = lsn; stuck++; break; _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
