Adds WATCHMAN_DISABLE_APM option #115
Conversation
|
@xfxf Thanks for the suggestion! The first thing that comes to mind is whether it makes sense to call New Relic out specifically, or should we have a more general New Relic is likely the most popular, but it could be nice to have a place to slot in support for removing from Datadog APM, AppDynamics, etc. without adding new settings for each. What do you think? |
|
Agreed, but I don't use the other APM tools, so I wouldn't be able to easily test this. |
|
@xfxf Yep, that's fine with me to just start with NR (under a more generic name). |
|
@mwarkentin done, see commit above. |
|
Thanks! I’ll try to take a more in depth look later today! |
mwarkentin
changed the title
|
@xfxf can you resolve the conflicts with master? |
There was a problem hiding this comment.
@xfxf Hey, sorry about the delay! Left a few small comments on the docs / code.
If you could add an item to the history file as well that'd be great!
# Conflicts: # AUTHORS.rst # watchman/views.py
|
@mwarkentin done. |
|
Merged, thanks! I have one more update I want to include in the next release and then I'll push it up to pypi! |
I've added a WATCHMAN_NEWRELIC_IGNORE option, which when enabled, disables instrumentation in the New Relic agent (used in many Django production installations) for any watchman calls.
This was done because I'm using watchman with ECS (on AWS) to monitor containers, and having watchman being hit multiple times every 30 seconds was causing a lot of New Relic stats that are based on averages to report totally inaccurately (throughput, apdex score, transaction times, etc) as the watchman queries end up being dominant.
There's no test, but I'm not clear how exactly I could test this outside of making the newrelic agent a dependency (which we don't want) or mocking the library completely (which makes me question what we're testing). The code is written defensively.