[Linux-ha-dev] [ANNOUNCE] Interim heartbeat packages refreshed
(2.1.2-4)
Andrew Beekhof
beekhof at gmail.com
Fri Sep 28 02:46:54 MDT 2007
this arguably deserves a public statement, even if it is in the form
of a "top 10" list.
1) the full version strings make a pretty clear distinction as to
which is which:
x.y.z vs. x.y.z-a.b
2) such suffixes are common on most distros
3) they are not release candidates - i feel very strongly on this point
4) as a wise man once said: broken is far worse than might-be-broken
5) running older versions is orders of magnitude worse than any
interim release and
causes just as much "confusion" on the lists as 2.1.2-4
6) the feature vs. bugfix distinction is largely redundant as it's a
very long time since
I wrote a 'pure' feature... ie. one that wasn't addressing a bug
report or long-standing
user gripe
7) interim releases will continue to be put out once a month (you can
plan on it)
8) the underlying cause is arguably (there is that word again) that
parts of the project are
at very different points in their development cycles but are forced
to share the same
release schedule
9) there is no release schedule
10) i have no control over 9)
On 9/26/07, Max Hofer <max.hofer at apus.co.at> wrote:
>
> It seems history is reapting itself (see realease 2.1.0).
>
> I already see an upcoming confusion about release versioning again.
>
> Incoming bug reports against a version which was never release by the HA
> project. Who released what etc.
>
> I really understand that the Novel guys (who are currently the active
> developers) try release as often as possible - and it seems communication
> between them and the offcial release manager is not the best.
>
> But the current release management is a disaster. Appart from beeing a PITA to
> know what a release really brings it is a nightmare to plan for the future
> (which is quite important from us sysadmins).
>
> Classification between bug-fix releases and "new-feature"-releases are really
> important - specially in an HA environment.
>
> Why not using common standards used by the open source community?
>
> Which would mean the interim release would have been called 2.1.3-rc1,
> 2.1.3-rc2 and so on.
>
> Advantage:
> * people see from the rcX tag that it is not offcial
>
> * a cleaner separation of the two malng lists
> - people using rcX know they have to ask on the dev-maling list (because it is
> RC)
> - the normal user maling list can be used to help people with common HA
> problems
> - the confusion about the different version on the user-maling list is kept at
> a minimum
>
> * if Alan (or whoever has the authroirty) decided
> - to pull some bugfixes from -dev back to 2.1.2 he could release (whenever he
> wants) 2.1.2.X.
> - to freeze rcX for a release he can
>
> My two cents - but I know - we already had this discussion 2 months ago.
>
> kind regards Max
>
> On Wednesday 26 September 2007, Andrew Beekhof wrote:
> > Just a quick note to say that the packages at
> > http://software.opensuse.org/download/server:/ha-clustering
> > were refreshed today after sufficiently (see pending bugs below)
> > passing automated testing.
> >
> >
> > The most important addition from my perspective is the more intuitive
> > colocation mechanism.
> > More details on the changes (and generally how colocation works) can
> > be found at:
> > http://oss.beekhof.net/~beekhof/heartbeat/docs/Colocation-
> > Explained.pdf
> > and:
> > http://oss.beekhof.net/~beekhof/heartbeat/docs/Colocation-
> > Explained.html
> >
> >
> > The current version is 2.1.2-4.1 for .rpm and 2.1.2-4 for .deb
> >
> > The packages correspond to Heartbeat as of this revision:
> > http://hg.linux-ha.org/dev/shortlog/702e4f418ca8
> >
> > The source for this version can be obtained from:
> > http://hg.linux-ha.org/dev/archive/obs-2.1.2-4.tar.bz2
> >
> >
> > Some stats on the changes since the previous build (098c0993b1d0 /
> > 2.1.2-2):
> > * Statistics:
> > Changesets: 109
> > Diff: 188 files changed, 5926 insertions
> > (+), 2003 deletions(-)
> >
> >
> > Lazy paste from the change log...
> >
> > * Test hardware:
> > + 4-node VMware Server cluster (sles10-sp1/256Mb) on a single
> > host (opensuse10.2/2Gb/2.66Ghz Quad Core2)
> > + 6-node EMC Centera cluster (sles10/512Mb/2Ghz Xeon)
> >
> > * Pending bugs encountered during testing:
> > 1377 - llm_get_nodename: index(1719751538) out of range
> > 1561 - Cluster node c001n03 returning after partition
> > 1562 - hostcache file rename error at startup
> > 1700 - HA Signon always reports success
> > 1711 - Shutdown blocked by call to cl_flush_logs()
> > 1712 - Shutdown blocked by ha_logd
> > 1714 - Stonithd core dumps
> >
> > * Changes since 2.1.2-2
> > + High: Core: Prevent logging buffer overflow - ensure the entity
> > is NULL terminated. Patch from Simon Graham.
> > + High: DTD: Remove stray character that made the DTD invalid
> > + High: PE: Ensure mandatory ordering constraints behave correctly
> > + High: PE: Implement a smarter algorithm for colocation
> > + High: PE: Include preferences of colocated resources when
> > promoting masters
> > + High: PE: Make sure only the correct number of cloned groups are
> > started
> > + High: PE: Make sure the merge_weights function is called for
> > complex resources
> > + High: PE: More accurate method of detecting colocation loops in
> > native_merge_weights
> > + High: PE: Optimize the merge_weights functionfor groups
> > + High: RA: Remove bashism from Filesystem OCF agent
> > + High: RA: Filesystem - Fix path reference.
> > + High: RA: o2cb - Fix path broken in recent cleanups
> > + High: clplumbing: Use the correct environment variable when
> > deciding to log via ha_logd
> > + High: crmd: Tell the LRM to refresh the resource's default
> > parameters
> > + High: core: Bug 1697 - really remove proctrack sources
> > (performance/memory leak fix)
> > + High: core: An open call with the O_CREAT flag was missing the
> > file mode.
> > + Medium: Core: Ensure heartbeat flushes all its logs before exiting
> > + Medium: ipc: Up MAXMSG to match MAXUNCOMPRESSED to allow larger
> > clusters.
> > + Medium: Logging: Sanitize and centrally define the log facility
> > used by various subsystems
> > + Medium: PE: Bug 1705 - Don't segfault when allocating empty groups
> > + Medium: PE: Bug 1710 - A resource's failcount was ignored on
> > nodes with no operations for the resource
> > + Medium: PE: Enact a saner default for rsc_order.score (s/0/
> > INFINITY)
> > + Medium: RA: EvmsSCC - Handle start failures caused by peers
> > starting at the same time (Patch from Jo De Baer)
> > + Medium: RA: Treat migrating (status 1) as running to avoid -
> > patch by Per Andreas Buer
> > + Medium: Stonith: Remove code-reuse to prevent excessive logging
> > by the LRM
> > + Medium: Stonith: ipmilan: fix potential segvfault in options
> > parsing
> > + Medium: cib: Improved transparency when in operating in degraded
> > mode
> > + Low: Admin: Bug 1691 - crm_resource doesn't complain about empty
> > strings as property values
> > + Low: build: configure.in: fix for platforms with libc in /lib/tls
> > + Low: CIB: Allow notification to remote clients
> > + Low: CRM: DTD - Expand the list of accepted values for boolean
> > fields
> > + Low: CRM: DTD - Explicitly specify a default for
> > rsc_colocation.symmetrical
> > + Low: DTD: Include helpful comment for rsc_order constraints
> > + Low: PE: Downgrade logging when we fallback to default option
> > values
> > + Low: PE: More efficient logging of orphan resources
> > + Low: PE: Prevent users from trying to use on_fail=fence and
> > stonith-enabled=false
> > + Low: PE: Stop the master placement score from affecting the
> > synapse's priority
> > + Low: RA: Add a tomcat OCF agent from Yamauchi Hideo
> > + Low: RA: ManageVE - Fix status when VE files are not persistent
> > + Low: RA: Put in updates to the Informix script as supplied by
> > Lars Forseth
> > + Low: RA: apache - drop annoying messages if no configuration
> > files are found
> > + Low: Stonith: external/ssh - simplify the reset command and wait
> > the same amount of time as the non-external variant
> > + Low: Stonith: Disable ipmilan by default
> > + Low: TE: Pause for a moment before trying to reconnect to stonithd
> > + Low: Tools: Allow pingd to change identity based on the name of
> > its executable - workaround for Bug #1701
> > + Low: logd: Update the default config file to reflect the new
> > default log facility
> > + Low: tools: haresources2cib.py: update parameter lists and
> > configuration defaults
> > * Unclassified impact
> > + lrm: Bug 1458 - include timing information in the op data.
> > + lrm: Bug 1684 - a meta-data operation fix introducing short
> > sleep on zero reads
> > + lrm: log non-recurring operations
> > + lrm: performance - replace linear with binary search for msg type.
> > + lrm: performance - remove unnecessary memstat
> > + lrm: raexecocf: a fix for bug introduced in 2a787bbd73f6; longer
> > sleep in get_resource_meta
> > + RA: Bug 1692 - monitoring fails for apache OCF resource. Put
> > in simple fix to apache resource agent monitoring from Sebastian
> > Reitenbach
> > + RA: eDir88 - fix to allow meta-data to work in case NDS is not
> > installed
> > + RA: pgsql - patch to remove bashism
> > + Stonith: Put in an APCmastersnmp.c patch from Peter Kruse which
> > makes the plugin work with new APC master switches
> > + Stonith: Bug 1615 - get the list of hosts from an RA asynchronously
> > + mgmtd: allow old clients to connect
> >
> > _______________________________________________________
> > Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> > Home Page: http://linux-ha.org/
>
>
>
> --
> Max Hofer
> APUS Software G.m.b.H.
> A-8074 Raaba, Bahnhofstraße 1/1
> T| +43 316 401629 11
> F| +43 316 401629 9
> W| www.apus.co.at
> E| max.hofer at apus.co.at
> _______________________________________________________
> Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> Home Page: http://linux-ha.org/
>
More information about the Linux-HA-Dev
mailing list