SGI
Open Source
Linux FailSafe FAQ / Known Bugs and Workarounds

If your question isn't answered here, feel free to ask it on the mailing list!

What is Linux FailSafe?
Should I be saying 'yes' to the question "Preparing to delete database at /var/cluster/cdb/cdb.db" during install time?
On the GUI, why can't I see the updated status of the node once I start HA services on it?
Why does the GUI crash?


What is Linux FailSafe?

Linux FailSafe is a high availability solution for your business critical applications on Linux. Linux FailSafe provides a clustered HA application platform in software avoiding the overhead of specialized fault-tolerant hardware.

In the event of a failure, Linux FailSafe automatically fails over applications from one system in the cluster to another. In combination with shared disk storage a Linux FailSafe cluster provides resilience from any single point fo failure and acts as insurance against unplanned outages. Planned outages (upgrades, reconfigurations) can be staged with no downtime.

Because of Linux FailSafe's unique design your application doesn't need to be modified or even recompiled in order to become highly available.

Linux FailSafe and related high availability products are built on a common infrastructure and all code resides in user space.

Should I be saying 'yes' to the question "Preparing to delete database at /var/cluster/cdb/cdb.db" during install time?

Yes. If you do not do that, cmond will complain saying "cmond[nnn]: Could not open configuration database." in the logs.

On the GUI, why can't I see the updated status of a node or a cluster once I start HA services on it?

Each time you define a node or a cluster and start HA services on it via the GUI, the status still shows 'Unknown'. To resolve this you need to kill all cad processes on the node running the GUI server and restart the GUI. The cad process will be automatically started by cmond. You will then be able to see 'Online' status.

Note: The correct status for nodes and the cluster can always be accessed through cmgr, using the command 'show status'

     > show status of cluster 
     > show status of node 
     

Why does the GUI crash?

We are actively trying to resolve this issue. There could be inconsistencies in the Java Runtime Environment for Linux that we are trying to address at the moment.