It's annoying to have to switch systems in order to figure out what happened on a given ONI Agent job. NCA should capture this stuff, at least in cases of an error.
This kind of depends where we go with #347 : if we end up with a single job that both runs and waits for the agent, we attach logs to that new job. If we don't, we really want to attach logs to the "run" job, not the "wait" job, but that's not really doable since logs aren't available when the "run" job is done.
Maybe this is a vote for merging the two and figuring out how to make that work....
Also, if we update the agent to parse logs and return more specific information, we may be able to do away with this ticket entirely. A clearer response from the agent means we don't need the low-level logs (and ONI's logs are often not super helpful). So this is blocked by #347 for the moment.
It's annoying to have to switch systems in order to figure out what happened on a given ONI Agent job. NCA should capture this stuff, at least in cases of an error.
This kind of depends where we go with #347 : if we end up with a single job that both runs and waits for the agent, we attach logs to that new job. If we don't, we really want to attach logs to the "run" job, not the "wait" job, but that's not really doable since logs aren't available when the "run" job is done.
Maybe this is a vote for merging the two and figuring out how to make that work....
Also, if we update the agent to parse logs and return more specific information, we may be able to do away with this ticket entirely. A clearer response from the agent means we don't need the low-level logs (and ONI's logs are often not super helpful). So this is blocked by #347 for the moment.