Comment 5 for bug 1291991

I attempted to apply the proposed patch by Dan Prince into the Icehouse codebase that still uses Nova Baremetal, and it did not solve the issue. I was still running into the issue when using DynamicLoopingCall, but only against certain DRAC firmware.

DRAC 7 1.40.40 - Not affected
DRAC 7 1.45.45 - Not affected
DRAC 7 1.57.57 - Affected, even after applying patch

I have attached a shell script that triggers the bug. This is an attempt to duplicate what happens with DynamicLoopingCall, although my understanding of DynamicLoopingCall may be incorrect (does it double the backoff time, or just increase it?) It is called with:
$ ./power.sh <username> <password> <ipaddr> off

I only ran into this bug when turning the hosts off, they were less sensitive when powering on. When powering off, this shell script (which introduces a 1 second longer delay each time) will trigger the bug about 2 out of 3 runs.

My concern is that we may not have fixed the issue by changing from FixedIntervalLoopingCall to DynamicLoopingCall in Ironic, if the bug is still triggered sometimes. Perhaps we need logic to accept the "insufficient resources" error as transient, and retry rather than error out.