Environment: TSM 5.5.2 2 x Novell OES 10.2 ( based on SLES 10) servers: WIS-HB32 and WIS-HB31 Novell Cluster Services 1.8.4 One NSS-formatted volume /media/nss/D31 is SAN-attached to both servers, but mounted on WIS-HB31 only. When WIS-HB31 goes down Novell cluster is capable of migrating and mounting this volume on WIS-HB32.
I've installed and configured the BA client v5.5.2 as described in Appendix B of "BA Client Installation and User's Linux Guide 5.5" which talks about configuring BA client in a cluster environment. Both clients are configured with the same NODENAME WIS-HB and share dsm.opt file that reside on the shared NSS volume in /media/nss/D31/tsm/dsm.opt along with the password file. The problem: Initial backup of the NSS volume executed as NODENAME WIS-HB on WIS-HB31 works fine and backup of all new data is performed. Subsequent incremental backup run on WIS-HB31 also works as expected (backing up only tiny amount of changes). However, after the same NSS volume is migrated to WIS-HB32, the incremental backup executed as NODENAME WIS-HB from WIS-HB32 sends all of the data to the server like if it was new data that was never backed up before. Anyone has any experience with such setup and/or have any idea why incremental of the same filespace from the second node, under same NODENAME behaves like selected backup? -- Warm regards, Michael Green