` #2158

New issue

Open

opened 2026-03-01 17:56:55 +03:00 by kerem · 1 comment

kerem commented

2026-03-01 17:56:55 +03:00

Owner

Originally created by @matthazinski on GitHub (Oct 8, 2022).
Original GitHub issue: https://github.com/ArchiveBox/ArchiveBox/issues/1035

I noticed listing snapshots is really slow. I found a few optimizations that seem to increase performance, but wanted to know whether there any undesirable side effects to this:

get_dir_size() requires recursing the snapshot directories for every snapshot on a page. It seems archive_size() caches this, but only for the default Django cache TTL (300 seconds). Is there any reason we can't just set CACHES['default']['TIMEOUT'] = None to ensure these keys don't expire by default?
Archivebox doesn't expose any options for choosing an external cache, which isn't great when running in ephemeral containers. I've had luck with configuring django_redis in settings.py so the cache can be periodically written to disk. If there's interest, I can put up a PR which optionally enables this. (Alternatively, if there are no blockers to upgrading to Django>=4.0, we can use the built-in Redis cache client.)
With a warm cache, Snapshot.from_json() requires a round trip to both the DB and the cache. Calling self.tags_str() with nocache=False seems to cut-down DB latency by about half according to the Django debug toolbar.

Originally created by @matthazinski on GitHub (Oct 8, 2022). Original GitHub issue: https://github.com/ArchiveBox/ArchiveBox/issues/1035 I noticed listing snapshots is really slow. I found a few optimizations that seem to increase performance, but wanted to know whether there any undesirable side effects to this: - `get_dir_size()` requires recursing the snapshot directories for every snapshot on a page. It seems `archive_size()` caches this, but only for the default Django cache TTL (300 seconds). Is there any reason we can't just set `CACHES['default']['TIMEOUT'] = None` to ensure these keys don't expire by default? - Archivebox doesn't expose any options for choosing an external cache, which isn't great when running in ephemeral containers. I've had luck with configuring `django_redis` in `settings.py` so the cache can be periodically written to disk. If there's interest, I can put up a PR which optionally enables this. (Alternatively, if there are no blockers to upgrading to Django>=4.0, we can use the built-in Redis cache client.) - With a warm cache, `Snapshot.from_json()` requires a round trip to both the DB and the cache. Calling `self.tags_str()` with `nocache=False` seems to cut-down DB latency by about half according to the Django debug toolbar.

kerem commented

2026-03-01 17:56:56 +03:00

Author

Owner

@pirate commented on GitHub (Nov 18, 2022):

Note that our DB latency is almost nothing as we use SQLite which serves from memory + filesystem, so adding cache for DB queries (e.g. from_json) is not worth it. But cache is still useful to avoid heavy filesystem operations like get_dir_size.

If you want to open a PR for that I'd review it!

I don't want to add redis by default / push people towards redis, it's too much complexity for now, but happy to add a filesystem cache, computed db field, or in-memory cache that doesn't require additional config to setup.

@pirate commented on GitHub (Nov 18, 2022): Note that our DB latency is almost nothing as we use SQLite which serves from memory + filesystem, so adding cache for DB queries (e.g. `from_json`) is not worth it. But cache is still useful to avoid heavy filesystem operations like `get_dir_size`. If you want to open a PR for that I'd review it! I don't want to add redis by default / push people towards redis, it's too much complexity for now, but happy to add a filesystem cache, computed db field, or in-memory cache that doesn't require additional config to setup.

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

starred/ArchiveBox#2158

No description provided.

Rows
Columns

[GH-ISSUE #1035] Cache get_dir_size() to avoid slow list rendering performance on /admin/core/snapshot/ #2158

[GH-ISSUE #1035] Cache `get_dir_size()` to avoid slow list rendering performance on `/admin/core/snapshot/` #2158