Indicating state of clustering

Is there any way of indicating to a user the state of system clustering i.e. whether a machine has failed, what machine is currently the master etc. ?

At the moment a brief dialog appears while the system is changing from the master to a backup node, and nothing at all appears when reverting to the master. (Also, when reverting to the master DB SQLTags remain at the old (backup) values until they are updated.)

We should add some system SQLTags to this effect.

As for the DB tags not updating: the value of the tags should be the same on both nodes!

Sorry, I should have pointed out that I have MySQL running on both nodes, with the Standby node replicating the Master node. This means that both nodes will be the same until the Master fails (I am disconnecting it from the network), after which any changes made through FactoryPMI are made to the Standby database.

I know that reconnecting the Master will then require the MySQL databases to be brought back into synch manually, but I noticed that data stored in other (non-SQLtags) tables is displayed from the Master, while SQLTags values (DB tags at any rate) stay at the Standby values until they are changed.

Yes, that is how SQLTags works - tags only get updated on change. You’re going to have to implement some sort of workaround for your quasi-clustered database. As part of your procedure for bringing up the master database, you can artificially cause the tags to think they’ve been changed by running a query like:

UPDATE sqlt_core SET valuechange = CURRENT_TIMESTAMP

You can also try using two-way replication for MySQL.

Yes, I would see standard replication with MySQL as being a bit of a stop-gap solution. If it happens things continue on (albeit without historical data) but you will definitely need to carry out a manual recovery process. If the process is important enough it probably warrants a full clustering solution at the database level.

I haven’t tried 2-way replication with MySQL - I’ll let you know if I get this working.