zmq: (don't) re-implement task message retries?

[oliver-sanders]

See also #2978

Simple feature removed in the initial Python3 port. For more information see:

$ git grep TASK_MSG_MAX_TRIES
$ git grep 3331

[oliver-sanders]

Result of offline discussion with @matthewrmshin: maybe we shouldn't do this? The retry will almost always fail anyway.

[hjoliver]

Yeah I recall a discussion (with @matthewrmshin) of this somewhere else recently (but I don't recall where). I tend to agree - message retries are almost always a waste of time, so we shouldn't bother. Polling (job query) is the only effective way to catch/recover-from status messaging failures.

[hjoliver]

@oliver-sanders - can you check that I've got everything (in PR #3402) ... your grep 3331 instruction above doesn't make sense as this issue IS 3331 😬

[matthewrmshin]

Re: the conversation on removing task message retries was this comment:
#3348 (comment)

[oliver-sanders]

@oliver-sanders - can you check that I've got everything (in PR #3402) ... your grep 3331 instruction above doesn't make sense as this issue IS 3331 grimacing

I think I added a TODO in the code with "see #3331", seems to have been removed by another change.