Network Error in filter drop-downs

[ukeswani]

Bug description

We have had superset deployed via the helm chart in various environments, with the same chart values in each environment, running ok for the last few months. Yesterday we started to see Network error for a number of, but not all, filter drop-downs in dashboards in just one environment. Refreshing the page the Network error's are random, not appearing for the same filters each time or all filters but always some. The errors are immediate, rather than appearing after some time. We see the same thing in different dashboards. The filter is querying a Trino dataset, and when we check in Trino, the queries for those error'd filters do not show - ie. it looks like those queries did not make it to Trino. We're not seeing any issues with the Trino pods.
In the superset logs we see this ValueError: bytes is not a 16-char string error correlating to the times we see errors in the dashboard. We do not see these errors in other environments (which do not have this issue).

Does anyone have any ideas what could be happening?

Screenshots/recordings

Superset version

master / latest-dev

Python version

I don't know

Node version

I don't know

Browser

Chrome

Additional context

apache/superset:GHA-dev-17114760203

psycopg2-binary==2.9.6
trino[sqlalchemy]==0.321.0
sqlalchemy-trino==0.5.0
Authlib==1.6.4

Checklist

• I have searched Superset docs and Slack and didn't find a solution to my problem.
• I have searched the GitHub issue tracker and didn't find a similar bug report.
• I have checked Superset's logs for errors and if I found a relevant Python stacktrace, I included it here as text in the "additional context" section.

[dosubot]

Hi @ukeswani! I'm Dosu and I’m helping the superset team.

This issue is almost certainly caused by malformed UUIDs in your Superset metadata database. Superset expects UUID fields (like those on slices, dashboards, or filter state) to be 16-byte binary values. If a filter or slice references a UUID that isn't valid (for example, a string that's not 16 bytes), SQLAlchemy will immediately raise a ValueError: bytes is not a 16-char string before any query is sent to Trino, which matches your symptoms: immediate "Network error" in the UI and no query reaching Trino (source).

This can happen if there was a recent migration, import, or data corruption event in the affected environment. The randomness is likely due to which filter or dashboard references a bad UUID at any given time.

To resolve:

1. Check your Superset metadata DB (e.g., Postgres/MySQL) for malformed UUIDs in tables like slices, dashboards, and any filter state tables. Look for UUIDs that are not 16 bytes or are obviously invalid.
2. Compare the affected environment's metadata with a healthy one to spot differences.
3. Fix or regenerate any malformed UUIDs you find (e.g., update them with valid UUIDs).
4. After fixing, restart Superset and verify that the errors are gone.

If you want to dig deeper, examine your Superset logs for the exact UUID values causing the error, then trace those back to the corresponding records in your metadata DB.

Let me know if you want example SQL queries for finding malformed UUIDs or more details on how to repair them.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}