I am not sure this will be an issue or not. As the EXITSTATUS of the jobs are 0, I assume that the jobs finish successfully. These condor_write failures usually happened at the end of life cycle of the condor jobs. Could you shed some light on what could be the reason that caused these condor_write failures?
It's probably just the shadow exiting. If the jobs succeed, there's probably nothing to worry about, but it wouldn't hurt to check the corresponding shadow log to make sure it isn't complaining about anything.
- ToddM