[druid] Catch IOError when fetching Druid datasource time boundary #3897
Conversation
|
The result of this query is used to determine when to look for metadata. I don't think the method should just return |
|
Also think we should fall back on current date time. I think that used to be the default before people said they had issues with past or future datasets. You could probably get in the git history to see how it used to be. |
john-bodley
force-pushed
the
john-bodley-druid-datasource-time-boundary-fix
branch
from
3621d7f
to
7004738
Compare
|
@Mogball I updated the logic to revert to the current time. @mistercrunch I went through the code to where this was added, and it seems like there's never been a check for this, i.e., here's the code from 0.8.5. |
superset/connectors/druid/models.py
Outdated
| max_time = dparse(max_time) | ||
| try: | ||
| results = client.time_boundary(datasource=self.datasource_name) | ||
| if not results: |
There was a problem hiding this comment.
Just noticed this now but we could fall back to datetime.now() when the function returns None as well
There was a problem hiding this comment.
@Mogball I updated the logic per your comment. PTAL.
john-bodley
force-pushed
the
john-bodley-druid-datasource-time-boundary-fix
branch
from
7004738
to
eb8fed4
Compare
mistercrunch
added
🏷️ bot
Whilst debugging #3894 I noticed that the calling the PyDruid client
time_boundary(...)function, anIOErrorcan be thrown as described here.This PR ensures we catch the exception and simply return
Nonewhich is consistent with the case that no results are obtained.to: @mistercrunch @Mogball @xrmx