Closed (fixed)
Project:
Apps
Version:
7.x-1.x-dev
Component:
Code
Priority:
Normal
Category:
Bug report
Assigned:
Unassigned
Reporter:
Created:
12 Jan 2012 at 19:11 UTC
Updated:
23 Mar 2012 at 23:00 UTC
Jump to comment: Most recent file
Looks like caching was never finished on the manifest. When the internet connection is slow this creates a very slow process of navigating the apps pages as it queries and downloads everything on each page refresh.
| Comment | File | Size | Author |
|---|---|---|---|
| #6 | apps-manifest_caching-1403030.patch | 965 bytes | TomDude48 |
| #5 | apps-manifest_caching-1403030.patch | 966 bytes | TomDude48 |
| #1 | apps.caching.1403030.patch | 641 bytes | randallknutson |
Comments
Comment #1
randallknutson commentedHere is a patch that enables caching. This significantly speeds things up for me.
Comment #2
e2thex commentedI think all we need to do is remove $manifest_cache = false;
Comment #3
e2thex commentedThis has been address in http://drupalcode.org/project/apps.git/commit/13ef0ca
Comment #5
TomDude48 commentedI think the changes in #3 are different from the original issue. The manifests are still not caching. I have attached a patch to re enable manifest caching.
This greatly speeds up Apps page loading times and reduces the requests to the server.
Comment #6
TomDude48 commentedOpps, sorry there was a bug. This one is correct.
Comment #7
febbraro commentedDo we really want to cache for 24 hours?
Comment #8
TomDude48 commentedWhere you thinking more or less? I thought that was a good balance but I am open to a different amount of time.
One concern is how slow the apps page loads are right now for the end user. Any amount of caching, even if it is just an hour will go a long way to solving that. (It has already made my testing significantly faster)
My concern longer term is how much stress it is putting on the Apps server. We currently have people hitting the server 100 times a day or more. There are a few times where the server has gotten overloaded. We are already looking at bumping up the instance size of that server to deal with the request we currently have. Hopefully in a few months it will be 100x that. ;-)
Comment #9
randallknutson commentedKeep in mind that they can always clear cache to get a new manifest from the server so a longer cache time is not necessarily bad.
Comment #10
febbraro commentedIf I'm not mistaken, the cache clear (like drush cc all) would only clear things that are cached as temporary. The timestamp will keep it in the cache for at least that long, after which it becomes temporary.
Comment #11
febbraro commentedThanks for pointing it out guys. I put caching in place, but for now left it at 1 hour TTL.