fix: add retries for .taskcluster.yml fetches #30
Conversation
bhearsum
force-pushed
the
pr-policy-failure
branch
4 times, most recently
from
5e68a86 to
db1cd30
Compare
bhearsum
changed the title
bhearsum
force-pushed
the
pr-policy-failure
branch
from
db1cd30 to
abe6f8e
Compare
| raise e | ||
| client = RetryClient( | ||
| client_session=aiohttp_session(), | ||
| retry_options=ExponentialRetry(attempts=5, statuses={404}), |
There was a problem hiding this comment.
It seems odd to me that a 404 would be intermittent; are we sure that's what's happening?
We should probably retry on 5xx though, if we're adding retries...
There was a problem hiding this comment.
It seems odd to me that a 404 would be intermittent; are we sure that's what's happening?
As far as I can tell, the only way we can ever end up returning None from tcyml.get is by hitting this line, which can only happen on a 404.
We should probably retry on 5xx though, if we're adding retries...
Somewhat confusingly, 5xx responses are retried even when statuses are passed:
Important: by default all 5xx responses are retried + statuses you specified as statuses param If you will pass retry_all_server_errors=False than you can manually set what 5xx errors to retry.
I'll add a comment to this effect.
bhearsum
force-pushed
the
pr-policy-failure
branch
2 times, most recently
from
d9e3fd2 to
abfab52
Compare
We get this intermittently for whatever reason, and it's constantly busting deployments.
bhearsum
force-pushed
the
pr-policy-failure
branch
from
abfab52 to
6b078bd
Compare
We get this intermittently for whatever reason, and it's constantly busting deployments.
I tried to debug the 404s a bit but didn't get anywhere; I don't think it's worth digging into more unless we see it with retries.