Zephyr "CoAP Bad Request" During Upgrade

Just did a release and it is failing after starting the download with the message <wrn> golioth: Server reports CoAP Bad Request. (Check payload formatting).

Just starting to debug it, but hopefully someone else has seen this before. This with Zephyr 3.5 and Golioth v0.8.0. Last upgrade was 3 weeks ago, so seems to be something changed between then and now.

[00:00:09.648,000] <inf> golioth_system: Starting connect
[00:00:10.779,000] <inf> bms_golioth: Starting Golioth firmware upgrade thread, our version is 0.1.1+0000000000
[00:00:10.780,000] <inf> golioth_system: Client connected!
[00:00:11.089,000] <inf> bms_golioth: DFU - Starting firmware upgrade from 0.1.1+0000000000 to 0.1.2+2024021601
[00:00:11.490,000] <wrn> golioth: Server reports CoAP Bad Request. (Check payload formatting)
[00:00:11.490,000] <err> bms_golioth: Error while receiving FW data: -14
[00:00:11.490,000] <inf> golioth: cancel and free req: 0x3fccce88


Thanks for reporting this Eric! We’re looking into this issue

1 Like

A quick update, I was able to upgrade to a previous release without issue, so it seems to be just the latest release. The only thing different is that I had added a tag to “Release Flags” for the latest release and have removed that even though it should not affect the upgrade. Still see the issue.

I will try creating a new release and see if that changes anything.

Deleting and Recreating the release didn’t help.

However, deleting and reuploading the artifact (same hash was shown) and deleting and recreating the release solved the problem.

Not sure what the root cause was, but happy I don’t have to drive 1.5 hours to the test site to do a manual upgrade :smile:

Would love to know what the root cause was.


1 Like

@EricNRS thank you for this detailed report! Could you share when the initial artifact was uploaded?

The initial artifact was uploaded on Feb 16 around noon New Zealand time, so Feb 15 around 2 PM Pacific. I did use it in a release with a Release Flag since it was a debug build which I had not done before.

I actually did two releases on that day with the first one around 8:15 AM New Zealand time and this was (and still continues to be) fine.

@EricNRS thanks for that info! We had a brief incident related to our storage provider in the window during which the original artifact was uploaded, which resulted in persistence of the artifact record, but not the underlying content. This issue has been resolved, which is why re-uploading your artifact addressed the error reported by your firmware. Apologies for interruption!

1 Like

Awesome, thank you for confirming what happened.

1 Like