Commit Graph

18 Commits (139f7646a7e3af491d5d6c6a99e82d46f9551400)

Author SHA1 Message Date
Eugen Rochko 5d3feed191
Refactor fetching of remote resources (#11251) 2019-07-10 18:59:28 +02:00
Eugen Rochko 406b46395d
Fix URLs appearing twice in errors of ActivityPub::DeliveryWorker (#11231) 2019-07-07 03:37:01 +02:00
Eugen Rochko bc60d794f8
Change ActivityPub::DeliveryWorker to not retry HTTP 501 errors (#11233) 2019-07-02 00:59:53 +02:00
Eugen Rochko 0d9ffe56fb
Add request pool to improve delivery performance (#10353)
* Add request pool to improve delivery performance

Fix #7909

* Ensure connection is closed when exception interrupts execution

* Remove Timeout#timeout from socket connection

* Fix infinite retrial loop on HTTP::ConnectionError

* Close sockets on failure, reduce idle time to 90 seconds

* Add MAX_REQUEST_POOL_SIZE option to limit concurrent connections to the same server

* Use a shared pool size, 512 by default, to stay below open file limit

* Add some tests

* Add more tests

* Reduce MAX_IDLE_TIME from 90 to 30 seconds, reap every 30 seconds

* Use a shared pool that returns preferred connection but re-purposes other ones when needed

* Fix wrong connection being returned on subsequent calls within the same thread

* Reduce mutex calls on flushes from 2 to 1 and add test for reaping
2019-07-02 00:34:38 +02:00
ThibG 9efcca3c54 Retry ActivityPub inbox delivery on HTTP 401 and 408 errors (#10812)
HTTP 401 responses returned by Mastodon's inbox controller may
be temporary if, for instance, the requesting user's actor/key json
could not be retrieved in a timely fashion. This changes allow retries
instead of dropping the message entirely.

Also added HTTP 408 as that error is by nature temporary.
2019-05-23 15:00:30 +02:00
Eugen Rochko 11955600ad
Skip deliveries to inboxes that have already been marked as unavailable (#9358) 2018-11-27 19:15:08 +01:00
Eugen Rochko cabdbb7f9c
Add CLI task for rotating keys (#8466)
* If an Update is signed with known key, skip re-following procedure

Because it means the remote actor did *not* lose their database

* Add CLI method for rotating keys

    bin/tootctl accounts rotate [USERNAME]

Generates a new RSA key per account and sends out an Update activity
signed with the old key.

* Key rotation: Space out Update fan-outs every 5 minutes per 1000 accounts

* Skip suspended accounts in key rotation
2018-08-26 20:21:03 +02:00
takayamaki 587da93152 checking http status code with range (#7544) 2018-05-19 14:47:44 +02:00
Eugen Rochko 97f02f2c08
Do not raise delivery failure on 4xx errors, increase stoplight threshold (#7541)
* Do not raise delivery failure on 4xx errors, increase stoplight threshold

Stoplight failure threshold from 3 to 10
Status code 429 will raise a failure/get retried

* Oops
2018-05-19 00:23:19 +02:00
Eugen Rochko d4de2239b0
Add a circuit breaker for ActivityPub deliveries (#7053) 2018-04-07 21:36:58 +02:00
Akihiko Odaki 54b273bf99 Close http connection in perform method of Request class (#6889)
HTTP connections must be explicitly closed in many cases, and letting
perform method close connections makes its callers less redundant and
prevent them from forgetting to close connections.
2018-03-24 12:49:54 +01:00
ThibG a0de3222dd Retry delivering toots over ActivityPub for about 2 days (#6298)
Currently, Mastodon will retry delivering toots for a bit over 1 hour.
This is a very short timespan when considering private and direct toots, which
cannot be seen by the recipient at all after the delivery attempts have failed.

Ideally, private and direct toots should have a different number of retries,
but I do not know how to do that.
2018-01-19 15:49:48 +01:00
abcang 2eab41cd1a Close connection when succeeded posting (#5390)
* Close connection when succeeded posting

* Update webmock
2017-10-14 14:38:57 +02:00
ThibG f7c909e290 Retry ActivityPub delivery a few more times (#5014) 2017-09-30 16:01:46 +02:00
Eugen Rochko f4ca116ea8 After 7 days of repeated delivery failures, give up on inbox (#5131)
- A successful delivery cancels it out
- An incoming delivery from account of the inbox cancels it out
2017-09-29 03:16:20 +02:00
abcang 3d9b8847d2 Flush body when POST requests (#5128) 2017-09-28 15:04:32 +02:00
Daigo 3 Dango a0bbeafb04 Suppress backtrace when failed to communicate with a remote instance (#5076) 2017-09-24 11:14:06 +02:00
Eugen Rochko b7370ac8ba ActivityPub delivery (#4566)
* Deliver ActivityPub Like

* Deliver ActivityPub Undo-Like

* Deliver ActivityPub Create/Announce activities

* Deliver ActivityPub creates from mentions

* Deliver ActivityPub Block/Undo-Block

* Deliver ActivityPub Accept/Reject-Follow

* Deliver ActivityPub Undo-Follow

* Deliver ActivityPub Follow

* Deliver ActivityPub Delete activities

Incidentally fix #889

* Adjust BatchedRemoveStatusService for ActivityPub

* Add tests for ActivityPub workers

* Add tests for FollowService

* Add tests for FavouriteService, UnfollowService and PostStatusService

* Add tests for ReblogService, BlockService, UnblockService, ProcessMentionsService

* Add tests for AuthorizeFollowService, RejectFollowService, RemoveStatusService

* Add tests for BatchedRemoveStatusService

* Deliver updates to a local account to ActivityPub followers

* Minor adjustments
2017-08-13 00:44:41 +02:00