Sending messages to other servers is broken #9185
Comments
|
1.26.0.rc3 is not a version that exists, can you confirm what version you're using? Generally the "could not serialize access due to concurrent update" error isn't something to be concerned about see #4993. Did you have outgoing federation working and it broke or has it never worked? Additional logs of what the federation sender is doing would help. |
|
Sorry, I meant rc1… And federation sending used to work, it just broke today or so after working for months. |
We will not be able to help you without |
|
Oh, sure. Decided to use a pastebin because of the amount of output this generated… |
|
(I had some difficulty loading the commie.io link, so here's a github gist mirror just in case). |
anoadragon453
added
T-Other
|
You have to wait a moment (maybe 30 seconds or so) until it loads the text… But sure, appreciated. |
|
I also tried setting up a federation sender worker, which did not help fix the issue, and seems to roughly give me the same logs. If those are needed, let me know. |
|
I believe It's possible that this error isn't caught in the From your logs I see a lot of transactions succeeding, with Interestingly I don't see any transactions to matrix.org. Is that perhaps the other homeserver you were testing with? |
anoadragon453
added
X-Needs-Info
|
Me and different users of my server tested several different servers in total. I myself mainly tested with privacytools.io… Jonah (@jonaharagon), the admin of aragon.sh, also got an account on my server and performed some tests, coming to the conclusion that it "doesn't make any sense" due to the fact that events from my server were clearly showing up in the logs of his server, yet, no messages from it were visible nevertheless. |
|
Hmm, I can see transactions going through to aragon.sh in your logs. I can also see transactions from -- Check whether the event is marked as an outlier
select event_id, processed, outlier from events where event_id = '$xxx';
-- Check whether the event has been soft-failed
select event_id, internal_metadata from event_json where event_id = '$xxx';
-- Check for any rejected events from the homeserver
select * from rejections where event_id like '%fantasycookie17.onederfultech.com';To alleviate the original If that doesn't alleviate the issue, then I'm at a bit of a dead end. You could try PM'ing me from the affected account at |
|
Another error that appeared in my I raised |
|
That error looks like the worker config is broken. Can you share your config? |
|
Well, I would, however, it seems like my server works again, for whatever reason… Messages have been arriving at other servers… This also means the config can't be broken, as I have |
|
I think @auscompgeek means that your worker is unable to talk back to the main Synapse process due to a misconfiguration in the If that is misconfigured, then federation traffic can still be sent, but things may still break in subtle ways. |
|
Oh. I see. That's my |
|
Nope, that's ok. So I'm realising that this is probably something we should really bold in the worker documentation, but Edit: or we should at least shout loudly if one hasn't included |
|
Ah. My bad, I guess? Anyway, changed that now, and I'm not getting any errors anymore. |
|
Glad to hear, and we should really fail in a louder way if that happens. Most important thing is that it's working now 🙂 Closing this issue then. |
Description
Sending messages to other servers does not work, however, receiving works, and read events and typing events go through.
Steps to reproduce
Expected behaviour
The message should arrive.
Actual behaviour
It does not. Log says
Version information
Homeserver: fantasycookie17.onederfultech.com
Version: I first ran 1.25.0, but then was recommended to switch to 1.26.0.rc3 as I was encountering an error in the logs that was fixed by this.
Install method: 1.25.0 via XBPS, 1.26.0.rc3 manually via
git clone.Platform: Void Linux in a KVM on Fedora Server. x86-64.
The text was updated successfully, but these errors were encountered: