1328661 - Running command on one-off dyno produces not stdout output

Status: RESOLVED INVALID

(Tree Management :: Treeherder, defect)

Description

[Armen [:armenzg]]

I first run this:
heroku run --app treeherder-stage bash

and then run the command I want:
./manage.py analyze_failures

Unfortunately it does not print anything into stdout.

Comment 1

[Ed Morley [:emorley]]

What's the expected output?

Comment 2

[Armen [:armenzg]]

vagrant ~/treeherder $ ./manage.py analyze_failures
[2017-01-04 21:20:08,576] INFO [treeherder.seta.analyze_failures:103] failures: 1
[2017-01-04 21:20:08,577] INFO [treeherder.seta.runnable_jobs:46] Fetching https://index.taskcluster.net/v1/task/gecko.v2.mozilla-inbound.latest.firefox.decision
[2017-01-04 21:20:09,089] INFO [treeherder.seta.runnable_jobs:49] For mozilla-inbound we found the task id: JJZ2wntARC6_hA23JoUDXQ
...
and continues

Comment 3

[Ed Morley [:emorley]]

Ah so these are:
* log level INFO
* coming via the treeherder.* loggers

In production the log level for the treeherder.* loggers is ERROR and above:
https://github.com/mozilla/treeherder/blob/78f66bc7c5f831acd675ef0f7224ac07f3ec8889/treeherder/config/settings.py#L178-L181

Compared to DEBUG and above in Vagrant (where DEBUG is set):
https://github.com/mozilla/treeherder/blob/78f66bc7c5f831acd675ef0f7224ac07f3ec8889/treeherder/config/settings.py#L216-L220

So this isn't Heroku-specific, and is working as configured at the moment, though open to ideas for adjustments to log levels across different parts of our infra. (There are some changes planned in bug 1318021, though changing the prod log level wasn't part of the plan so far).

Comment 4

[Ed Morley [:emorley]]

In your one-off dyno you could `export TREEHERDER_DEBUG=1` prior to running the command to see all output, if that helps?