(PXP-10739): Add bucket region to DRS #355
Conversation
|
The style in this PR agrees with This formatting comment was generated automatically by a script in uc-cdis/wool. |
Avantol13
left a comment
Avantol13
left a comment
There was a problem hiding this comment.
This caching strategy isn't super useful when there are many instances of the service running (since all the caches are in-memory and specific to that pod). Since we expect many instances of indexd to be running, I would suggest we look into a more centralized caching solution similar to what's been done in Fence (look for how we dealt with caching creds during signed URL calls or even some of RAS stuff maybe). The idea being: we need a central place for newly spun up instances of indexd to get the cached information. Naive solution (and what we did in Fence) was have a new database table to store the central cache AND in-mem cache in the instances
indexd/index/drivers/alchemy.py
Outdated
| bucket_region_info = response.json() | ||
| cache.set("bucket_region_info", response.json(), timeout=3600) | ||
| else: | ||
| print( |
There was a problem hiding this comment.
Use the logger for this.
indexd/index/drivers/alchemy.py
Outdated
| ) | ||
| break | ||
| else: | ||
| print( |
tests/test_aliases_endpoints.py
Outdated
| @@ -1,3 +1,5 @@ | |||
| import os | |||
indexd/app.py
Outdated
|
|
||
| from alembic.config import main as alembic_main | ||
| import cdislogging | ||
| from distutils.command.config import config |
There was a problem hiding this comment.
Do you still need this config here?
Jira Ticket: PXP-xxxx
New Features
Breaking Changes
Bug Fixes
Improvements
Dependency updates
Deployment changes