-
-
Notifications
You must be signed in to change notification settings - Fork 678
federation issues and "This server is using an older version of Matrix. Upgrade to Matrix v1.5 to use Element without errors." #3304
Comments
Having the same issue using the docker version of dendrite, and taking the image down and restarting to get latest actually made my instance completely unreachable by Element. |
How can I check what Matrix version a server supports? Update: Is it this one?
Guess we have to pin now to older Element versions... |
This got introduced in
Element 1.11.53 is still fine. |
And the Dendrite source agrees: dendrite/clientapi/routing/routing.go Lines 140 to 149 in dae1ef2
Only up to v1.2 announced. |
Related context: conduit-rs announces up to v1.4, so that'll also trigger the warning: https://gitlab.com/famedly/conduit/-/blob/next/src/api/client_server/unversioned.rs?ref_type=heads#L28 Suggesting Synapse is the only(?) server that is not causing warnings in Element right now. |
@S7evinK I just updated to main, do I need to delete jetstream or something? Maybe is trying to work on all the late msg's and getting stuck?
is not always the same user this is just a snippet. Also this is interesting about the MSC goin to dissable it and see if it helps
[edit2]
|
At least all of the mentioned servers don't seem to really reply to requests at all and should be blacklisted at some point. (tested with https://federationtester.matrix.org/) And for the initial question: No, no need to delete Jetstream. It may be possible that there are thousands of messages in the streams already and are now being processed. So the interesting part would be right after start up, how many messages are being restored by NATS (most importantly for the |
@S7evinK to late, I was getting spam like 200 msg a sec from that same msg regarding user real-changeling so had to stop and do how we used to do when we first added jetstream, and deleted the folder ( we do not mind if we lose old msg's) now seems like the spam has stop, also have removed the MSC for the config and that is gone... local msg's are working good again, not sure about federation so far is not, going to let it sit for 1-2 hours. Regarding that msg about 8m for jetstream how can we fix/improve that? |
Unsure about the 8MB warning, we used to pump quite a bit of data (iirc, 20MB+) through Kafka before switching to JetStream, not sure how this looks at the moment. But this shouldn't be a problem. And if matrix.org is receiving your messages, I don't think your server is blocked or anything. |
same here, https://matrix.ari.lt/ |
@S7evinK I just check on the federation test thing, as domain hispagatos.org as usuall and says is having issues... we have not changed anything in our side in years regarding this, did we change the way we point a domain to a home server? our homeserver is matrix.hispagatos.org but our matrix domain is hispagatos.org
and also on the nginx server..
you can see the .well-known/matrix/server and client are present. |
Right, so: both well-known files should be served on https://hispagatos.org instead of https://matrix.hispagatos.org If you need to use SRV records, hispagatos.org needs a valid TLS certificate. (this is not recommended, though). So my course of action would be:
|
ok wait this rings a bell now! @S7evinK so hispagatos.org our blog where I also have those .well-known was down because of a DDOS attack for the whole week last week to sourcehut sr.ht ... and now they have some broken git pages ..... this now does make sense because... is the only thing really that has happened between last week and this week since it was working 100% fine. @S7evinK thanks like always I appreciate your time. and was able to fix the spamming I had looking into the logs etc so win win situation, the issue with the matrix version is still there but seems is not creating any issues to our server. |
Completely broken with latest |
Going to close this here, as there is currently nothing we can do - it is already reported to the Element Web team. |
This appears to be fixed in element nightly now |
Nice. Here's the "partial revert": matrix-org/matrix-js-sdk#4014 |
Updated to last stable git version of dendrite mid december
version: v0. 13.15 https://github.com/matrix-org/dendrite/releases/tag/v0.13.5
People is reporting they can't see msgs from matrix.org or even other homeservers (this was working last month just fine )
since all the report from my users, at least 6 of them for the last 2 weeks I decided to log using element ( I use iamb and gomuks )
and the first weird thing I notice is :
I have checked postgresql I see no errors
I need to enable debug for dendrite since the regular stdout I see no errors neither
posting this now so I do not forget, since I work a lot and always push things, and then I end up forgetting and my users keep asking me to fix this.
Hopefully this is a known issue and has a fast fix/workaround.
let me know what other info people will need when I enable this weekend debug ( I have a lot of work now ) I will post the results as well.
is possible we are block? how can we check that?
[edit] msgs work from our homeserver to matrix.org is the other way around, I had family members write to me and friends on matrix.org and I never got their msg's since Sunday. I saw them because I used my matrix.org test account and can see matrix.hispagatos.org msgs on matrix.org but not the other way around :(
[edit2] this is what I get to stdout
thats about it
The text was updated successfully, but these errors were encountered: