"No state group for unknown or outlier event" error in federation #13026
Comments
|
possibly related to #12507. |
|
This looks like a duplicate of #12507 to me. |
|
1.61.0 still has this issue. |
|
Also, the query |
|
not the same as #12507 then! |
|
In 1.61.0, I observe another exception that also emits "No state group for unknown or outlier event", don't know what happens. Some exception begins with "Exception: Error fetching missing prev_events for ...", and some StopIteration exception. Here's the log of ~2min since one start. |
erikjohnston
added
S-Minor
|
@mytbk Curious, are you using retention settings and/or https://github.com/matrix-org/rust-synapse-compress-state/? |
|
My retention settings: I also used rust-synapse-compress-state before. |
|
I may be in a similar situation: I have configured a default retention time of 240d on my server (and I run API: # curl --header "Authorization: Bearer XXX" 'https://matrix.uni-marburg.de/_synapse/admin/v1/rooms/!OGEhHVWSdvArJzumhm:matrix.org/forward_extremities' | jq .{
"count": 5,
"results": [
{
"event_id": "$w__T5VCKv34SVaoVF4LoIynZDo5QqJ4mgzCq9XCaaAg",
"state_group": 79056,
"depth": 466487,
"received_ts": 1656433481438
},
{
"event_id": "$FJU7WQNE9c1QcZYZvh5p9o_f1mnXBPuoYbccaCWZNbk",
"state_group": 80290,
"depth": 466977,
"received_ts": 1656600524544
},
{
"event_id": "$m9S8nYfhe-MBDODoxTR-33HRd8-jtt35KoVUV7AbGuQ",
"state_group": 79705,
"depth": 465281,
"received_ts": 1656529102606
},
{
"event_id": "$3uLkuz0dKwartpR4_qTPEKezBtp31a0goxBZXfnU9nw",
"state_group": 79711,
"depth": 463617,
"received_ts": 1656348839094
},
{
"event_id": "$X-1osxxtdtlvcre3QZEcIooCMl3qGE4epxL5ahiLQ8E",
"state_group": 79709,
"depth": 463617,
"received_ts": 1656348838539
}
]
}DB: The DB shows one entry more than the API i.e. Every event I receive for As you can see the event mentioned in the Traceback is the one extra forward extremity in the database. My working hyposthesis is, that bad things start to happen if you get forward extremities outside the retention window which may or may not have something to do with the use of |
|
First I tried increase the retention time which did not resolve my issue. But a got the room to sync again. |
|
I can confirm that got rooms for me working as well. |
|
I can see deleting from event_forward_extremities can improve this. |
|
Those who find deleting rows from @mytbk: I'm confused: first you said that you didn't have any rows in |
richvdh
added
the
X-Needs-Info
|
Not in my case. At one point I used an inner join in one of my queries to get additional information about the events in question. SELECT to_timestamp(e.origin_server_ts/1000),
e.*
FROM EVENTS AS e
JOIN event_forward_extremities efe ON e.event_id = efe.event_id
WHERE e.room_id = '!OGEhHVWSdvArJzumhm:matrix.org'
ORDER BY origin_server_ts DESC;The problematic event did exist in the |
|
@FrankNagel please can you open a separate issue explaining exactly what you did and what you know? It's very confusing to try to track several different problems with superficially related symptoms on a single issue. We can always merge them later if we realise they have a similar root cause. |
|
@richvdh The query in #12507 |
MadLittleMods
added
the
A-Corruption
Description
In synapse to 1.60.0, I found the "No state group for unknown or outlier event" run time error in the journal (in log output section).
There are a lot of such events, giving a runtime error about once per minute.
Steps to reproduce
Start the synapse service and view the log with journalctl, then see the runtime errors.
Homeserver
matrixim.cc
Synapse Version
{"server_version":"1.60.0","python_version":"3.10.5"}
Installation Method
Other (please mention below)
Platform
Arch Linux matrix-synapse package 1.60.0-1.
Relevant log output
Anything else that would be useful to know?
No response
The text was updated successfully, but these errors were encountered: