matrix.org/static/jira/browse/SYN-411

53 lines
1.9 KiB
Plaintext

---
summary: Should expose a way to force the HS to retry pending transactions to the AS
---
created: 2015-06-14 17:07:22.0
creator: neb
description: |-
Submitted by @matthew:matrix.org
See @mirw:matrix.mirw.cw-ngv.com's feedback from tadhack June 14
id: '11648'
key: SYN-411
number: '411'
priority: '2'
project: '10000'
reporter: neb
status: '1'
type: '2'
updated: 2016-11-07 18:28:07.0
votes: '0'
watches: '3'
workflowId: '11749'
---
actions:
- author: kegan
body: |-
The AS went down, so the HS would've been backing off. Matthew's instinct here was right : "i suspect it's doing exponential backoff on retrying to the AS".
I'm open to suggestions on how to improve this, probably providing a way to kick the timers to retry immediately. How the kick is exposed (http poke, shell command, etc) and whether this is spec worthy would make for an interesting discussion.
created: 2015-06-14 17:23:35.0
id: '11861'
issue: '11648'
type: comment
updateauthor: kegan
updated: 2015-06-14 17:23:35.0
- author: kegan
body: |-
For record keeping, the transcript from #matrix:
I've just hit an issue where synapse doesn't seem to be notifying my application service of events. I see the events going from my browser to synapse... but then not on to the application service. homeserver.log doesn't show synapse attempting to send notifications (and they don't show up in tcpdump either). This was working earlier, but then there was a period where my application service was down, and I wonder whether synapse has "blacklisted" it?
created: 2015-06-14 17:27:25.0
id: '11862'
issue: '11648'
type: comment
updateauthor: kegan
updated: 2015-06-14 17:27:25.0
- author: richvdh
body: 'Migrated to github: https://github.com/matrix-org/synapse/issues/1334'
created: 2016-11-07 18:28:07.0
id: '13655'
issue: '11648'
type: comment
updateauthor: richvdh
updated: 2016-11-07 18:28:07.0