I’ve recently built a complete SharePoint 2013 installation on Windows Azure consisting of 3 servers: SharePoint Server, Office Web Apps and SQL Server. No resilience built in, as it’s only for testing and R&D. I found that the farm kept experiencing timeout errors, random errors which couldn’t be cleared and then would suddenly just clear and everything would be just fine for a while. Sometimes the errors would vanish after a simple retry of a page, other times the faults seemed ‘hard’. Normally an IISREST or a server reboot would fix the issue, but having the farm randomly be unavailable was plainly not a good thing.

On examining the ULS logs many of the errors pointed at database timeouts and of the unavailability of the SQL Server, so I did what I suspect is an unsupported action and that is to increase the timeout of the database connection from the default of 15 seconds to 45 seconds. Voila! All the errors have disappeared. A little birdy told me that Microsoft are investigating this issue as they’ve have several reports of the problem.

One thing to mention is this business of unsuportability, whilst what I’ve done is not a supported action, I’ve learnt recently from a reliable source that Microsoft will only cease to support an installation if it can be ascertained that the unsupported action/configuration/metric is what is causing the issue. In this case the action, supported or not has actually cured the issue, so there should be no problem. Also it’s a test/dev environment rather than production. You make you’re own decision if you to are getting these random errors, I mention these things without prejudice …

Here’s the Registry Key to change by the way which contains the database connection string:

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Shared Tools\Web Server Extensions\15.0\Secure\ConfigDB

Cheers

Dave Mc