Strange ThreadPool size warning

mefoster

It seems that the *first* time I start an IceGridNode on my machine, it prints the following message:

10/02/07 09:49:42.694 warning: thread pool `IceGrid.Node.ThreadPool' is running low on threads
Size=1, SizeMax=10, SizeWarn=8

If I stop the node and restart it, then it comes up right away with no ThreadPool warning. If I stop and restart both icegridregistry and the node in question, then I usually get that warning again.

The warning doesn't make sense, does it? Why should it possibly be warning with those values for Size*? I'll attach my icegridnode config file in case there's something weird going on.

BTW, this is using Ice 3.2.1 on Fedora 7 i386, using the packages included in Fedora.

benoit

Hi,

Thanks for pointing this out!

This warning is expected if there are many servers registered with the node as the registries invoke on the node to load the servers. To disable the warning you can set the IceGrid.Node.ThreadPool.SizeWarn property to 0.

Alternatively, you can change the ServerEntry::sync call to ServerEntry::syncAndWait in src/IceGrid/NodeSessionI.cpp line 248. This will prevent the registries from sending too many concurrent requests to the node when it starts. We'll add this fix for the next release.

Cheers,
Benoit.