dcrcli error Error Locking the Mutex

[akemp]

I use homebrew scripting to cross validate the inventories of two autonomous LMS installations. Sometimes I get this error "Error Locking the Mutex" and the only way I can recover is to recycle the DCR server it occurs on.

Anyone have a clue what exactly this is ? I'm assuming the DCRcli is attempting to lock the DB for exclusive access so it doesnt get modified while its doing its operations, but cant find it documented anywhere (well except google in dev forums)

[Joe Clarke]

Is this the exact error? I cannot find this in the code.

[akemp]

Yes , verbatim as output from the CLI. it might be an error message from the underlying database and the way it handles assertions.

[Joe Clarke]

I found it. It doesn't come from the database, but rather from our regdaemon.xml accessor code. We're trying to lock a semaphore, and this is failing.

When this happens, it would be helpful to get the output of ipcs as well as a truss of the dcrcli process:

truss -a -f -vall -rall -wall -o truss.out dcrcli ...

This would need to be run as root.

[akemp]

I'll need to wait for it to occur again before I can capture it. Its an infrequent occurance so I'll keep track of it.