======================================================================
Halcyon PrimeAlert (R) High Availability
Version 1.3.0
Release Notes
======================================================================
Copyright (c) 1998-2008 Halcyon Monitoring Solutions, Inc.
Note: For the latest information on this product, visit:
http://www.HalcyonInc.com
Refer to the README file(s) for all other information.
Versions Documented: 1.3.0 1.2.0a 1.2.0
1.1.0a 1.1.0 1.0.0a 1.0.0
======================================================================
VERSION/DATE: 1.3.0/Sept 23 2010
BUG FIXES
----------------------------------------------------------------------
- More resilient to unexpected problems and higher reliability
of successful failovers.
- The Backup Event Manager Status and Backup Failover Type fields in
the HAClient module will be updated regularly even when a failover
is not underway.
ENHANCEMENTS
----------------------------------------------------------------------
- The SNMP Retries and Timeout module load parameters have been made
configurable. These values may be increased in order to provide
better service in a noisy loaded network environment.
- The context menu "List Agents" adhoc command has been replaced with
"List Agents (Local topo)" and "List Agents (Remote topo)" commands.
Instead of reading the list of all IP Addresses of all agents that
have ever talked to this server layer, they query the Sun MC topology
manager to build a list of all agents. Thus old decomissioned addresses
will not be used. The "Local" and "Remote" refer to whether the
command queries the local topology manager or the remote one that is
configured as the "Backup Server" in the HAServer module.
These commands use snmpv2u, and as such will work properly in
environments where snmpv1 been has been disabled.
The new commands will indicate what changes have occurred since their
last execution, as well as which hosts do not have the HAClient module
loaded.
UPGRADE STRATEGY
----------------------------------------------------------------------
- This release obsoletes all patches for PrimeAlert High Availability.
This release is functionally equivalent to Patch_4107_D on agent
hosts. If you already have that patch installed, you do not need
to install this version.
Installation of this release will automatically uninstall all prior
High Availability agent patches. You do not need to uninstall the
prior patches.
KNOWN PROBLEMS and IMPORTANT NOTES
----------------------------------------------------------------------
- When creating an agent update image using es-imagetool, the list of
"Supported OS-Arch combinations" will include i686-sun-Linux_2.6.
This architecture is in fact not supported by this release. Do not
include it in agent update images nor attempt to deploy High
Availability to Linux platforms.
- If you intend to change the IP Address of a Sun MC Server Layer
but not the hostname, you must use IP Addresses in the HAClient
module load parameters. This ensures that HAClient reseeds
agents against the new IP, instead of simply re-using cached
credentials. The latter is done in order to speed failover
completion and decreases the load on the Sun MC cfgserver daemon.
- The "disabled" failover type in the HAServer module is only
intended to be used for "scheduled maintenance" of a server layer,
during which time the agents whose "primary" is that server layer
should not switch away. They are allowed to switch back to this
server layer.
The HAClient module does not use the "failover type" of it's
backup server layer.
To ensure that agents do not fail back to their primary server
layer until you want them to, ensure that the primary server
layer's HAServer module remains in the "Forced" failover mode,
or remains off, or does not have the HAServer module loaded.
- If the HAClient module is edited while it is in the middle of
a failover, the client may not fail over properly. It may not fail
over properly in the future either. The symptom to look for is
the agent.log will be filled with entries stating that "a failover
is pending" however no failover will occur. The workaround is to
manually stop and re-start the agent.
======================================================================
VERSION/DATE: 1.2.0a/2008-04-30
ENHANCEMENTS
----------------------------------------------------------------------
- Added support for Sun MC 4.0.
BUG FIXES
----------------------------------------------------------------------
- None.
UPGRADE STRATEGY
----------------------------------------------------------------------
- This release is compatible with the previous version.
- Any patches/workarounds provided by Halcyon prior to this release
should be removed. Specifically check ESDIR/bin and ESDIR/cfg
for files named HALHAClient-* (where ESDIR is often /var/opt/SUNWsymon/)
KNOWN PROBLEMS
----------------------------------------------------------------------
- If the HAClient module is edited while it is in the middle of
a failover, the client may not fail over properly. It may not fail
over properly in the future either. The symptom to look for is
the agent.log will be filled with entries stating that "a failover
is pending" however no failover will occur. The workaround is to
manually stop and re-start the agent.
======================================================================
VERSION/DATE: 1.2.0/2007-2-9
ENHANCEMENTS
----------------------------------------------------------------------
- The failover process now works with Solaris 10, with Service
Management Facility (SMF) enabled.
- Added Solaris whole root zone and sparse root zone support.
- Added x86/x64 Solaris support.
BUG FIXES
----------------------------------------------------------------------
- Minor fixes in the documentations.
- If the seed for a server is changed in the HAClient module the
agent will be re-seeded upon next failover to that server.
Prior to this, PrimeAlert High Availability used backups of
authentication records, thus necessitating that an agent
be manually re-seeded should a server layer's seed ever change.
UPGRADE STRATEGY
----------------------------------------------------------------------
- This release is compatible with the previous version.
- Any patches/workarounds provided by Halcyon prior to this release
should be removed. Specifically check ESDIR/bin and ESDIR/cfg
for files named HALHAClient-* (where ESDIR is often /var/opt/SUNWsymon/)
KNOWN PROBLEMS
----------------------------------------------------------------------
- If the HAClient module is edited while it is in the middle of
a failover, the client may not fail over properly. It may not fail
over properly in the future either. The symptom to look for is
the agent.log will be filled with entries stating that "a failover
is pending" however no failover will occur. The workaround is to
manually stop and re-start the agent.
======================================================================
VERSION/DATE: 1.1.0a/2005-3-28
Bug FIXES
----------------------------------------------------------------------
- Workaround a failure for some agents to properly exit during failover
UPGRADE STRATEGY
----------------------------------------------------------------------
- None. This release is compatible with the previous version.
KNOWN PROBLEMS
----------------------------------------------------------------------
- Same as 1.1.0
======================================================================
VERSION/DATE: 1.1.0/2004-10-17
ENHANCEMENTS
----------------------------------------------------------------------
- Invalid high availability configurations will be reported by the
modules. Examples:
a) the primary and backup server have the same ip address
b) the ip addresses of the servers cannot be resolved
- Added support for Sun Management Center's Agent Update feature.
BUG FIXES
----------------------------------------------------------------------
- Determine agent's server context based on ip address rather than
hostname.
- Module will install properly if Sun Managment Center is not
installed in the default directory.
- Fixed miscellaneous docmentation mistakes.
- On some systems, the agent is known to fork a temporary process when
it exits. The failover script will stop any forked processes so
that the failover process is not hindered.
- Regular expression matching has been improved so that failover will
not be affected by the hostname or size of the process id.
UPGRADE STRATEGY
----------------------------------------------------------------------
- None. This release is compatible with the previous version.
KNOWN PROBLEMS
----------------------------------------------------------------------
- Symptom:
The console reports the error message "Unable to obtain console info"
when opening the Host Details window for the agent.
Possible Causes:
- The server layer has not updated the agent with the security keys
used for communication. Wait and retry after five minutes.
- On a system with multiple agents (standard agent plus platform
agents), failover can result in an agent pointing to a server layer
with the incorrect security keys. This may occur if the agent is
restarted by the the user before the failover process has completed
but another agent has completed its failover steps.
- In rare occassions, if the user starts the agent just before the
failover process has completed, the agent will use the incorrect
keys.
Workaround:
1) Stop the agent
2) Remove the failover configuration
# rm -rf /var/opt/SUNWsymon/cfg/failovers
3) Reseed the agent
# /opt/SUNWsymon/sbin/es-run base-usm-seed.sh -s <secret_seed> -c <component>
where <secret_seed> is the seed used to initially configure
the agent and <component> is the name of agent being reseeded.
For example, the following command will reseed the default agent:
# /opt/SUNWsymon/sbin/es-run base-usm-seed.sh -s maplesyru -c agent
The following example will reseed the platform agent with the
instance name "platform1":
# /opt/SUNWsymon/sbin/es-run base-usm-seed.sh -s maplesyru -c platform1
4) Start the agent
======================================================================
VERSION/DATE: 1.0.0a/2003-07-29
ENHANCEMENTS
----------------------------------------------------------------------
- Added Sun MC 3.5 support.
UPGRADE STRATEGY
----------------------------------------------------------------------
- None. This release is compatible with the previous version.
======================================================================
VERSION/DATE: 1.0.0/2003-05-12
ENHANCEMENTS
----------------------------------------------------------------------
- None. This is a FCS release.
BUG FIXES
----------------------------------------------------------------------
- None. This is a FCS release.
UPGRADE STRATEGY
----------------------------------------------------------------------
- None. This is a FCS release.
KNOWN PROBLEMS
----------------------------------------------------------------------
- If a failover gets initiated while /tmp is full, the agent is not
able to restart.
Workaround:
Monitor the /tmp directory and ensure it never reaches capacity.
---//---