I am trying to figure out how TS may work in Veritas Cluster with automatic failover.
TS 6.1, OD 6.0 Solaris 8, Veritas cluster - 2 nodes.
From a previous
post and documentation I gather the following:
- The HA watchdog does NOT monitor TeamSite in a cluster environment. So, I am inclined to think that if using clustering, its better not to use HA. Instead use custom Agents to monitor TeamSite that do the failover.
- Install TeamSite on both nodes.
- Both nodes share one disk which is used for backing store.
- Configure Virtual IPs for the cluster.
- Create a custom agent which will monitor TS processes and failsover to secondary node in case of failure.
- Install OpenDeploy on both nodes. OpenDeploy has to be bound to real host IPs.
- If OD Scheduler is used, the schedules have to be replicated (how ?) to OD on both the nodes.
- The state of any running external tasks during failover is lost.
- Any custom scripts that parse the log files or rely on other temp files created on TS Server, may not produce desired results on failover.
Questions:
1. Is my understanding (above) correct ?
2. Anybody's got a custom agenat for failover that they can share ?
3. What processes should the custom agent monitor ? What I have so far is : iwserver, iwopenapi, iwproxy, iw-webd, servletd.
4. How to (best practice to) synchronize configurations between the two nodes - particularly server specific files such as iw.cfg, odbase.xml, etc.
5. Can OD on both the nodes point to same DB for reporting ? How about OD Scheduler DB ? (similar to TS pointing to same store)
6. In case of abnormal shutdown during some active TS operations, what happens to data in TS cache ?
Any other thoughts/comments ?
