chore(api): disable legacy calibration accessors on flex #13513
Conversation
These messages about missing calibrations are completely meaningless on the flex and are incredibly spammy; pull them. They might have meaning on the OT-2 but honestly we have a _lot_ of ui surrounding them so whatever.
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## chore_release-7.0.0 #13513 +/- ##
=======================================================
- Coverage 71.36% 71.35% -0.01%
=======================================================
Files 2418 2418
Lines 67937 67933 -4
Branches 7886 7886
=======================================================
- Hits 48480 48476 -4
Misses 17616 17616
Partials 1841 1841
Flags with carried forward coverage won't be shown. Click here to find out more.
|
| log.warning( | ||
| log.debug( |
There was a problem hiding this comment.
I might leave this as a warning, unless this is actually more false-positive than true-positive in practice.
There was a problem hiding this comment.
this is, unfortunately, a frequent occurrence when a 6.3.1 app is viewing a flex because it's polling the old pipette offset route. which i might want to just disable in the server or something
| log.warning( | ||
| log.debug( | ||
| f"Tip length calibration is malformed for {tiprack} on {pipette_id}" |
There was a problem hiding this comment.
I might leave this as a warning, unless this is actually more false-positive than true-positive in practice.
There was a problem hiding this comment.
this just never should ever print ever so i can make it a warning
Make the legacy routes for tip_length and pipette_offset 503 on flex and not try and get the offsets using the old calibration accessors, since they are hardcoded to use the ot2 versions of calibration which do not apply to the flex and dumped errors in the logs all the time. the api side change supports testing.
sfoster1
changed the title
| async def get_all_pipette_offset_calibrations( | ||
| pipette_id: Optional[str] = None, mount: Optional[pip_models.MountType] = None | ||
| pipette_id: Optional[str] = None, | ||
| mount: Optional[pip_models.MountType] = None, | ||
| _: API = Depends(get_ot2_hardware), |
There was a problem hiding this comment.
To make sure I understand correctly: on a Flex, even retrieving pipette calibrations through these endpoints was doing the wrong thing?
There was a problem hiding this comment.
Yes. This code was hardcoded to select the ot2 pipette offset loaders.
We had a lot of spammy logs in the api log about missing or malformed pipette and tip length calibration.
The missing calibrations should not be logged at info level - these are polled and this is a level state, so it just ends up being spammy. The malformed calibrations though should be, and it's weird that we were getting them.
The culprit is that <7.0.0 apps poll routes to get pipette offset and tip length because that's what you do on the ot2, and the implementations of those routes hardcode the ot2 calibration models - reasonable, since we completely replaced them for the flex. So you'd only get those logs if an old app was polling the robot.
To fix that, I made the routes shortcircuit with a 403 forbidden / not allowed on hardware error. This won't affect the flex because the app doesn't check that route on flex.