Serviceguard Version A.11.16 Release Notes, 2nd Edition, September 2004

Serviceguard Version A.11.16 Release Notes, Second Edition
Patches and Fixes in this Version
Chapter 144
coordinator node does not send an SMN down trap. The SNM
package current node MIB variable is also incorrect in the SNMP
MIB table.
• What was the resolution? Enhancements allow the subagent and its
API to identify SMN packages and treat them differently than
regular packages.
JAGae61889: cmGetsatus for packages returns -10 after online
reconfiguration.
• What is the problem? When packages are added or deleted
concurrently, cmsnmpd is trying to retrieve a status of a package
that has been deleted. This logfile output is seen:
cmGetstatus (CM_PDKG_STATUS) returns: -10 (errno=2)
***Error: retrieveing package statuses: -10
• What is the resolution? Added logging returns the package name not
found, and logs the errorno from cln_get_health().
JAGae61376 config commands (cmcheckconf and cmapplyconf) are
slow when not using the -k
• What was the problem? This is a follow-on defect to previous
Enhancement Requests submitted concerning the slow speed of the
SG configuration commands: cmquerycl, cmcheckconf and
cmapplyconf. These enhancements had been added to SG A.11.15 by
adding a new option -K to the configuration commands to bypass all
disks that are not part of the cluster lock. Using this new option
makes the configuration commands much faster. However, a second
part of the enhancements had not been provided in SG A.11.15.
This is to make the configuration commands faster when doing full
disk probing. As described in the previous Enhancement Requests,
there are two major areas where the ability to probe the disks in
parallel and streamline the opening and closing of the Volume
Groups to read the cluster attributes.
• What was the resolution? Config library now queries a few disks at a
time and can connect to multiple cmclconfds on a node to improve
performance. Changes made in config library and cmclconfd to
improve feedback to the user during disk data gathering, to remove
redundant opens and closes of devices, and to have multiple
cmclconfd processes gathering data at the same time.