PDA

View Full Version : Heartbeat checking when server is down



narraway
01-20-2011, 05:28 AM
We've seen some odd, but repeatable behaviour with heartbeat checking when the server goes down.

After the server goes down the heartbeat checks start to take more time. I have configured manual heartbeats to look more closely at this, but it happens with automatic checking as well. Initially the time taken by the call to lc_heartbeat doesn't change, but within a couple of calls it starts taking longer and the duration changes. So, for example, in one case I might see the first long duration as 18 seconds followed by 213 seconds and then 400seconds. Subsequent calls then take around 400 seconds. This is not only odd it's inconvenient too, since it makes it hard to give warnings to the user at consistent intervals. The durations of the calls seem to be dependent on the number of licensed features, which further complicates consistent, informative warnings to the user.

Anyone else experiencing this, or anyone know what might be going on?

(I've seen the problem on WinXP, Server 2003 and Server 2008 with libraries 11.4 and 11.9)