Attachment 152170 Details for Bug 234318 – List of changes since initial beta

List of changes since initial beta

rhel4_5_changes_from_beta.txt (text/plain), 2.26 KB, created by Jonathan Earl Brassow on 2007-04-10 18:48:53 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Jonathan Earl Brassow

Created: 2007-04-10 18:48:53 UTC

Size: 2.26 KB

patch

obsolete

>Cluster mirror changes since RHEL4.5 beta1 (cmirror-kernel-*2.6.9-25.0)
>=======================================================================
>** NEWEST
>
>** Removed overzealous BUG() statement in complete_resync_work.
>The case is valid if a region is recovering and has write request
>waiting for recovery to complete.  Only able to hit this bug if
>the mirror was being resync'ed and the secondary device is failed.
>
>** During server relocation (which can happen due to machine failure
>or normal mirror suspension), the server value could get set before
>the client had a chance to clean-up.  This caused the server to
>become confused and issue a BUG().
>
>** perform a flush of the log before suspending.  This ensures
>that regions which are in-sync get correctly flushed to the disk
>log.  Without this, there will always be recovery work to be done
>when a mirror starts up - even if it was properly in-sync during
>shutdown.
>
>** clean-up memory used to record region users when a mirror is
>shutdown.  It was possible for some regions to be left over
>(causing a memory leak) during certain fault scenarios.
>
>** properly initialize the state field (ru_rw) in the region
>user structure when a mark occurs.  Without the initialization,
>it was sometimes possible for the region to be misinterpretted
>as recovering instead of marked.
>
>** resolve and unhandled case in server_complete_resync_work
>
>** reset a variable in cluster_complete_resync_work.  Failure
>to do so was causing a retry to include the wrong value for
>the completion of the resync work - confusing the server.
>
>** Issue mark and clear region requests when the log is flushed,
>instead of independently.  This allows for locking code clean-up
>as well as following the rules of a mirror log more closely.
>Also adds 20% performance improvement on mirror writes.
>
>** Add sequence numbers to log requests.  It was possible under
>heavy load for requests to timeout and be resent.  Without the
>sequence number, the answers to the old requests were not
>properly discarded.  This caused client/server communication to
>be incoherent.
>
>** Obviate is_remote_recovering() function.  It did not cover
>all cases and the condition is better checked by the flush
>function.  Removal of the extra function call adds 20%
>performance improvement on mirror writes.
>
>** OLDEST

Actions: View

Attachments on bug 234318: 151276 | 151620 | 152170