You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We originally had s3fs interface to s3 object storage in this project.
It was replaced with the lower-level boto3in this PR - for consistency with other projects / developer familiarity. It meant a one-line call to s3.ls() was replaced with an involved key listing method borrowed from SO
Now (with 10x more data but still only thousands of objects) it's returning very slowly, with s3.ls() still taking seconds - you can see an observable lag in each call to the paginator.
Is it worth digging into why (e.g. the possibility of request throttling at the JASMIN end, because it feels like this behaviour has changed?) or take the simplest course just to return to s3fs? I don't have expertise in s3
The text was updated successfully, but these errors were encountered:
metazool
changed the title
s3 paginator key listing runs very slow
boto3 paginator key listing runs very slow
Jan 29, 2025
We originally had
s3fs
interface to s3 object storage in this project.It was replaced with the lower-level
boto3
in this PR - for consistency with other projects / developer familiarity. It meant a one-line call tos3.ls()
was replaced with an involved key listing method borrowed from SONow (with 10x more data but still only thousands of objects) it's returning very slowly, with
s3.ls()
still taking seconds - you can see an observable lag in each call to the paginator.Is it worth digging into why (e.g. the possibility of request throttling at the JASMIN end, because it feels like this behaviour has changed?) or take the simplest course just to return to
s3fs
? I don't have expertise in s3The text was updated successfully, but these errors were encountered: