Server Locks and the ViewX Experience

[aosburn]

As systems get more complicated, and components get added to databases, the amount of troubleshooting regarding server lock times continues to eat up more and more time. We do our best to optimize databases, but there's one constant complaint which arises from server locking: ViewX application hanging.

During the period of time where ViewX is not receiving updates from the server (due to networking or server locks), ViewX generally becomes unresponsive, and a 'grey' semi-transparent overlay is applied to the application by Windows. Windows then forces the application to close if (not when) a user continues to click at the display, trying to force a response. This will then cause Windows to force close ViewX (often referred to as ViewX crashing).

(Obviously not Geo SCADA, but the experience is the same)

My suggestion is that ViewX needs to be more robust and patient in regards to server timeouts. If the server is not responding, that's fine - but the application should have its own overlay. Particularly helpful would be a message saying "here is the current time, your server is not responding, contact your administrator". Even a cell phone picture then could help the admins track down the time and place of the offending lock.

But at the moment, users tend to attribute these locks to all sorts of things - and describe the 'grey screen' in a surprisingly diverse fashion. I feel that having ViewX handle these locks and indicate which server/node ViewX is pending a response from (if possible) would be a great step forward in resolving these issues.

[BevanWeiss]

Generally server locks and ViewX performance are independent.

I don't think I've really seen these correlated together.  Do you have logs or similar that show these to be connected?

ViewX certainly has its share of issues however, GeoSCADA Expert 2019 seems to have quite a few annoying issues with crashes, and alarm banner losing connection to systems etc.  But these are generally independent of the servers themselves.

The handling of multi-node systems does seem to be poorer in GeoSCADA Expert than previous ClearSCADA releases.  But this has typically only been an issue I've seen when actually failing over between servers, not just when a server is under load.

If you are regularly getting server lock alarms, then it's entirely likely that your servers are failing over (or going Main-Main) in a hot standby arrangement.  Otherwise your ViewX might be failing between servers based on the configured timeout settings.  Best solution here is to prevent your servers from encountering database locks that inhibit them from providing service.

[sbeadle]

Hi Aaron,

Thanks for the feedback. ViewX is aimed at being asynchronous to long transactions, and the product team have changed/improved this in many areas over the years. If you can report this to support they can feed this in. They will need client and server logs and a (UTC) time of the stall.

Thanks

Steve