Commit graph

40 commits

Author SHA1 Message Date
Samuel Clay
4f35c2fade Making feed.min_to_decay reflect the actual decay time. Using it in statistics, and refreshing it in feed fetching. Also splitting up feed munin. 2010-12-23 13:29:31 -05:00
Samuel Clay
97ad4838f1 Fixing stories in tabs, and adding option to specify username for refresh_feeds, to assist in development. 2010-12-14 23:10:13 -05:00
Samuel Clay
fa663e910a Faster feed refreshing (insta-fetching) by only computing the score for the current user and leaving the other users with dirty subscriptions. 2010-11-10 18:04:17 -05:00
Samuel Clay
4cf9b05f0c Changing all datetime.datetime.now() to datetime.datetime.utcnow(). 2010-10-10 23:55:00 -04:00
Samuel Clay
50488db7ca Close the connection for every set of tasks. 2010-10-08 16:08:41 -04:00
Samuel Clay
39d9e13449 Forcing refresh of feed information at every stage of feed fetch pipeline. This really needs a rewrite. Someday. 2010-10-03 18:04:40 -04:00
Samuel Clay
15a27e840f Bootstrapping mongo to use story_guids as unqiue to feeds. 2010-09-08 18:30:46 -07:00
Samuel Clay
41be0abde3 Moving feeds out of the queue when they get updated. 2010-08-31 22:04:26 -04:00
Samuel Clay
25c0450020 Commenting out map/reduce statistics because mongo keeps crashing. 2010-08-30 13:33:29 -04:00
Samuel Clay
abe73a1d0e Barring the ability to use multiprocessing for feed fetching, in the meantime we'll have to use separate processes, which means potential for fetchin gthe same feeds. 2010-08-23 08:09:58 -04:00
Samuel Clay
ac53f33788 Further Mongo work. Stories are now saved correctly, with tags + authors. Still need to do UserStories. 2010-08-21 20:42:38 -04:00
Samuel Clay
8b9c28e8f8 Cleaning up exception handling of feeds to show to users during import. 2010-08-18 21:56:30 -04:00
Samuel Clay
68b0ae7254 Adding exception handling UI. Also disabled feeds that are just chock full of exceptions. No need to continually fetch buggy feeds. Still need to figure out a way to let users fix problematic feeds, but letting them know is an improvement. 2010-08-18 20:35:45 -04:00
Samuel Clay
55683fc1d5 Skipping feeds with fewer than X stories_per_month. Used to catch the queue up and toss inactive feeds. 2010-08-16 13:21:32 -04:00
Samuel Clay
588207384c Logging tweaks to handle console and file handlers. 2010-08-16 12:52:39 -04:00
Samuel Clay
3c6fdfdea7 Statistics on all feeds, stories, averages. Shown in the new statistics modal (not yet finished). 2010-07-25 23:13:27 -04:00
Samuel Clay
2c6433b0de Parallelizing feed fetcher now recognizes other feed fetchers so they do not replicate work. Now multi-threaded and multi-processed aware 2010-04-27 13:44:53 -04:00
Samuel Clay
0a664b44a8 Adding feed update history. Gotta get some metrics on the speed of the biggest cpu/network suck. 2010-04-23 21:19:19 -04:00
Samuel Clay
71b746a48d Feed update cascading. Feeds update relative to how often they have new content. This should save a LOT of processing power. Whew. 2010-04-19 12:09:04 -04:00
Samuel Clay
f169569710 Adding single threaded feed updating (used for tests and debugging) 2010-04-09 16:37:19 -04:00
Samuel Clay
391aa76dd6 Lowering worker threads from 8 to 4 to conserve memory. 2010-03-02 17:04:54 +00:00
samuelclay
d9c7ae7771 Auto-compressing all stories and feeds at the db level for better database lookups. 2010-01-27 16:29:11 -05:00
Samuel Clay
7b08770b3a Getting rid of Queues entirely. Replaced with values being distributed directly to Processes. One day this will change. Until then, all feeds fit in memory. 2009-09-16 03:54:33 +00:00
Samuel Clay
a7888c2aae Switching from threads to processes. Sure, no big deal. Django is not thread-safe, so I can't use threads. I can, however, use multiprocessing to pass a Queue around between Processes. 2009-09-16 02:22:27 +00:00
Samuel Clay
0446ff2ee3 Adding locks to feed processing. 2009-09-12 20:42:38 +00:00
Samuel Clay
bec2f00979 Exiting after it's over. 2009-09-10 04:24:51 +00:00
Samuel Clay
0b9b778382 Queues not working. Reverting back to 1 feed at a time. Ugh. 2009-09-10 04:16:50 +00:00
Samuel Clay
8437ef29f1 Handling empty queues and increasing thread count 2009-09-10 03:52:43 +00:00
Samuel Clay
67ab0d6c66 Let's try changing the load. 2009-09-10 03:48:22 +00:00
Samuel Clay
02a4d292e8 Moving from threadpool to a Queue. Much simpler. 2009-09-10 03:33:05 +00:00
Samuel Clay
0caf49e9a4 OK. Turning on threading for feed fetching. Enough logging to catch the bugs. Fixed the segmentation fault on exit. 2009-09-10 02:31:55 +00:00
Samuel Clay
9e522ca528 Woo - Checking for content and title differences to resolve duplicate entries. Integration tests prove a number of differences are successfully resolved. 2009-08-30 00:43:13 +00:00
Samuel Clay
bdd91d714d Refactored the feed fetcher. Added integration tests for dupe matching. Also stubbed in AI score calculations so I can do something with the output soon. 2009-08-29 19:34:42 +00:00
Samuel Clay
d24b56fbf4 Adding new "Return to Page" button. When the user visits other pages in a feed, security prevents NB from reading the iframe, so we provide a convenient hook to bring the iframe back. Call it SnapBack, except, don't. 2009-08-26 02:31:35 +00:00
Samuel Clay
ce9ec58d00 Some feeds don't have pages. Who knew. 2009-08-23 00:07:37 +00:00
Samuel Clay
9c522416a1 Gotta get rid of threading until I can figure out how to thread-safe the Django ORM. It's corrupting my tables! 2009-08-23 00:00:55 +00:00
Samuel Clay
fa123e7f3d Updating feeds to crash with noise. 2009-08-21 13:14:44 +00:00
Samuel Clay
6cf04497ee Checking for threadpool 2009-08-20 02:53:41 +00:00
Samuel Clay
8545e09d9a Using threadpool module and feedjack to import feeds. Largely the same functionality, but now more debugging info, as well as better error handling and threads. 2009-08-20 02:43:01 +00:00
Samuel Clay
88ccfa7e77 Created admin commands to run refresh_feed/s without invoking Apache. Finally! 2009-08-14 01:48:21 +00:00