- Issued:
- 2006-03-27
- Updated:
- 2006-03-27
RHBA-2006:0196 - Bug Fix Advisory
Synopsis
clumanager bug fix update
Type/Severity
Bug Fix Advisory
Topic
Updated Red Hat Cluster Manager packages that fix various bugs and provide
enhancements are now available for Red Hat Cluster Suite on Red Hat
Enterprise Linux 3.
Description
Red Hat Cluster Manager provides high availability of critical server
applications in the event of planned or unplanned system downtime.
Various bugs in Red Hat Cluster Manager have been discovered and addressed
since the release of Red Hat Enterprise Linux 3, including:
- Device ordering not honored within a service
- lsof -b should be -bn to avoid blocking
- clusvcadm segfault if sent a command with services locked
- poor signal handling in cluquorumd
- clulockd hang causes infinite timeout loop
- Miscellaneous cluster locking performance problems
- Incorrect notation for clumembd%broadcast_primary_only flag
- Poor/unfriendly messages from clustat on members not in the quorum
- Not all services stopped if member leaves quorum during shutdown
Users of Red Hat Cluster Manager are advised to upgrade to these updated
packages, which resolve these issues.
Performing a Cluster Upgrade
This erratum can be upgraded in a "rolling" fashion, which means that it is
possible to upgrade the cluster software without taking the entire cluster
offline at once. Due to a bugfix between 1.2.3 and 1.2.9, however, it is
necessary to upgrade the lowest-ordered member last. (If you are upgrading
from 1.2.9 or later, you may upgrade the members in any order; the
instructions below assume you are running clumanager version 1.2.3.)
Notes (Before you begin):
- All cluster members must be running and participating in the cluster
quorum, or no members should be running at all.
- An ideal time to perform this upgrade would be during schedule
maintenance and/or when it will impact the fewest number of users.
- Switching to or from the netlink-based IP management utilities (available
in versions 1.2.27 and later) has a significant impact on the way
services are handled, and must be done with all cluster services
disabled, but with all cluster nodes participating in the cluster
quorum. Refer to the cludb man page for more information.
- Disabling ARP checking is only recommended for users who see "Not in
subnet" errors. This feature requires each node to be restarted, and
may require manual copying of /etc/cluster.xml to each member.
Refer to the cludb man page for more information.
Steps (with all cluster members running):
(1) Run the following command:
# cludb -g members%member0%name
The host displayed is the lowest-ordered cluster member. This member must
be upgraded *last*.
If this member is "not found", open the /etc/cluster.xml file and look for
the "member" tag with the lowest "id" property; this is the lowest-ordered
member. Make a note of this member; this member must be upgraded last.
(2) Run 'rpm -Fvh [filenames]' (as noted above) on all cluster members
except the member noted in step (1), one at a time. Do not execute this
command on multiple members simultaneously, as the upgrade process disables
a member temporarily (removing it from the cluster quorum). If too many
members are disabled simultaneously, the cluster quorum will no longer
exist, forcing services managed by the cluster software to become
unavailable.
Services located on a member that is being upgraded will migrate to other
cluster members whenever possible.
(3) Run 'rpm -Uvh [filenames]' on the member noted in step (1).
(4) If desired, balance cluster services across available members using
redhat-config-cluster or clusvcadm.
(5) Run 'chkconfig --add clumanager' on all members.
Steps (No cluster members running the cluster software):
(1) Run 'service clumanager stop' on all members.
(2) Run 'rpm -Uvh [filenames]' on all members.
(3) Run 'chkconfig --add clumanager' on all members.
(4) Restart the cluster software on all members if desired
Solution
Before applying this update, make sure all previously released errata
relevant to your system have been applied.
To update all RPMs for your particular architecture, run:
rpm -Fvh [filenames]
where [filenames] is a list of the RPMs you wish to upgrade. Only those
RPMs which are currently installed will be updated. Those RPMs which are
not installed but included in the list will not be updated. Note that you
can also use wildcards (*.rpm) if your current directory *only* contains
the desired RPMs.
Please note that this update is also available via Red Hat Network. Many
people find this an easier way to apply updates. To use Red Hat Network,
launch the Red Hat Update Agent with the following command:
up2date
This will start an interactive process that will result in the appropriate
RPMs being upgraded on your system.
If up2date fails to connect to Red Hat Network due to SSL
Certificate Errors, you need to install a version of the
up2date client with an updated certificate. The latest version of
up2date is available from the Red Hat FTP site and may also be
downloaded directly from the RHN website:
Affected Products
- Red Hat Enterprise Linux High Availability for x86_64 3 x86_64
- Red Hat Enterprise Linux High Availability for x86_64 3 ia64
- Red Hat Enterprise Linux High Availability for x86_64 3 i386
Fixes
- BZ - 171637 - lsof -b still blocks on gethostbyname causing slow force unmount; use -bn instead
- BZ - 172886 - Poor signal handling in cluquorumd
- BZ - 172893 - clulockd hang causes infinite timeout loop from connecting clients
- BZ - 172894 - Misc. cluster locking performance + bugfixes
CVEs
(none)
References
(none)
The Red Hat security contact is secalert@redhat.com. More contact details at https://access.redhat.com/security/team/contact/.