ArcMap crashing when overwriting services

[PeterTimmers]

Has anyone else had this experience?

I go to overwrite an existing map service with some small changes to the mxd (just the scale dependency changes) and ArcMap crashes.  The only way to deal with it is usually to delete the service and republish.

Happens every now and then and hard to pin down.

[ErinBrimhall · Accepted solution]

Has anyone else had this experience?

I go to overwrite an existing map service with some small changes to the mxd (just the scale dependency changes) and ArcMap crashes.  The only way to deal with it is usually to delete the service and republish.

Happens every now and then and hard to pin down.

Peter,

There have been other rumblings on this forum about problems with overwriting services.  See this thread.

It doesn't sound like the problem will be fixed until a later release or service pack, but the last post in the aforementioned thread describes a possible workaround.

[PeterTimmers]

Thanks for that.

I seem to get no issues with this problem on our dev server where it's a single machine with 4 cores instead of our production machine which is four virtual boxes with 2 cores each.

An associated problem is where I often get one or more of the 4 machines having trouble starting or stopping certain services - usually services that have a high usage.  Sometimes the only answer is to go through and stop and restart machines.

Anyway I'll try the various workarounds mentioned, like keeping an eye on the staging folder.

[ErinBrimhall]

Are the "Pooling" options of your services configured very differently between Production and the single, 4-core machine?  E.g. is Production configured with a much larger number of min/max instances?

If your services are running against an SDE Geodatabase on the back end you might consider using the "Administration" --> "Administer Geodatabase..." dialog new at 10.1 to cleanup any active connections/locks between your services and objects in the Geodatabase.  I've ran into situations where these connections were interfering with the ability to manipulate services.

[PeterTimmers]

We're currently running with a 9.3 SDE behind our 10.1.  Maybe that has something to do with it??

The maximum instances per machine would be different since our dev/test is a single 4 core while our production is 4 x 2 core machines.