In Jenkins we currently have the HTTP logging environment set to gather further information in case of update failures. Our mozharness script doesn't handle that right now, so it doesn't get created at the correct location beside all the other log files. I would suggest we set those env variables directly in the mozharness script and point it to the correct log location.
Thinking more about this bug and its implementation I'm coming more and more to the result that we might not want to have it directly in the mozharness script but keep it in our Jenkins code. If we would add it to mozharness we would need two options. One for the log levels and another one for the log file location. Whereby for the latter we could also assume `abs_log_dir`. So usually you would have to set environment variables like: > NSPR_LOG_MODULES=timestamp,nsHttp:3,nsHostResolver:5,append > NSPR_LOG_FILE=http.log</propertiesContent> Armen what do you think? Would such a feature be helpful? Not sure if the releng script would like to make use of it. If not, I would mark this bug as wontfix.
WONTFIX is fine. I don't even clearly understand what those logs you mention are :)
We discussed on IRC and agreed to close as incomplete given that it works for us when having this option set in our CI system. If Releng would need HTTP logs when running our update tests we can rethink about the implementation.