High Availability Monitors Version A.03.20 Release Notes

Chapter 1 13

Patches and Fixes in this Version

JAGac39442 When adding new physical volumes (PVs) to a volume group

(VG), the ServiceGuard that is monitoring the PV summary

gets halted (it fails over)

In order for PV summary to be monitored, there must be equal

numbers of physical volumes in each physical volume group

(PVG). Since physical volumes are added one at a time,

/etc/lvmpvg and /etc/lvmtab become out of sync, and the Disk

Monitor has no way of knowing that this is a temporary

condition.

The user should never change PVGs while the package is

running. Do not attempt to modify the underlying components

of a resource while you are monitoring that resource. Before

PVs are added to, or removed from, a PVG that is being

monitored, you must halt the ServiceGuard package, and then

restart it after the changes to the PVG are complete on all

nodes. See the example below:

cmhaltpkg <packagename>

vgchange -a e <volgroupname>

... make PVG changes ...

vgchange -a n <volgroupname>

cmrunpkg <packagename>

Killing the diskmond process is not recommended, since the

persistence client will detect its demise and restart it if there

are outstanding monitor requests.

Table 1-3 Corrected Defects

Defect number Problem and Resolution