"Failed to load timeline position" (/context 404) from jump to last unread message button after using Purge History API #16207
Comments
|
actually nevermind the exact even that the last unread message points to I can't seem to find this way? What i found before just had it listed in the event from !yHkWvIueFDsVBYAhRo:dmnd.sh{
"auth_events": [
"$L0qPnlPdup8ulMk4RfqhJvStgLDUBl4ngczvw6qYrY4",
"$YKavGK4PhIqWmMyFzpNu6aR1nHcha9w-cLiaNQB-FUI",
"$Zss-GH6gJXGRfZTjy2aDIjvc7nzVXOUkgAOWml090Fs",
"$-L8OOdBQgIQWQOVwjCYVjH3wr7MYbyn5dHg4E-8fTvA"
],
"prev_events": [
"$HvTnDTEDlze3Q15hditXICmEchw_ef7hehoKxx2O_Uk"
],
"type": "m.room.member",
"room_id": "!yHkWvIueFDsVBYAhRo:dmnd.sh",
"sender": "@daniel:dmnd.sh",
"content": {
"membership": "join",
"displayname": "Daniel",
"avatar_url": "mxc://dmnd.sh/eCrtVyqTKrlHahhCkxNFWhFX"
},
"depth": 23,
"prev_state": [],
"state_key": "@daniel:dmnd.sh",
"origin": "dmnd.sh",
"origin_server_ts": 1585755085445,
"hashes": {
"sha256": "ip1PpFOScC9EDnt4bvu3wmLMYJbrYT9p07KucmbHV/c"
},
"signatures": {
"dmnd.sh": {
"ed25519:a_Ysle": "7GAMP20JYJdQkYafud+Igrfv5OnHM4juAH42OJC38QTyqKQ4K+zwUvfunZFgSRkZAP7OlmFk+qiZdnRCHumWCg"
}
}
}still i can see plenty of messages (as in real messages with a body, not just join events etc) in all these channels in postgres that i can not see in my element scrollback on no device, nor can other people in these rooms. other event with a proper body{
"auth_events": [
[
"$1572119910866QDAAt:dmnd.sh",
{
"sha256": "sH4iQEw/dSfVo2XtXq1JMVX/3IZv2BcE5QLJNa0WMsc"
}
],
[
"$15550156812593YvpRC:dmnd.sh",
{
"sha256": "3UhmP24q+5eTGEebXPN6AmKnPVQvotDqrSjplDazvnM"
}
],
[
"$1572115012822ggxWZ:dmnd.sh",
{
"sha256": "e4sNcvvOljj6OKgqHcj2lJ6iHQmhTIeH6AXjfpJo7X0"
}
]
],
"prev_events": [
[
"$15757375341742LQAPQ:dmnd.sh",
{
"sha256": "uNAsG9nDcAuPQxTLxkErSoXB1/h+Hpg/Y/0qcnZoGw0"
}
]
],
"type": "m.room.message",
"room_id": "!MYczBHIFOSXqTpHVGB:dmnd.sh",
"sender": "@andre:dmnd.sh",
"content": {
"body": "hackerman.png",
"info": {
"size": 780459,
"mimetype": "image/png",
"thumbnail_info": {
"w": 610,
"h": 600,
"mimetype": "image/png",
"size": 570232
},
"w": 842,
"h": 828,
"thumbnail_url": "mxc://dmnd.sh/EYvdbKXiHuUIlDdkwJFkbtFS"
},
"msgtype": "m.image",
"url": "mxc://dmnd.sh/SntLwOugOttOyGqUDYMjSkSK"
},
"depth": 77,
"prev_state": [],
"event_id": "$15757380361757BgoMK:dmnd.sh",
"origin": "dmnd.sh",
"origin_server_ts": 1575738036831,
"hashes": {
"sha256": "oLNGBP6AojOXN4mWoWX8FHHDcPjMqdTH7UMC2MaN+Ks"
},
"signatures": {
"dmnd.sh": {
"ed25519:a_Ysle": "UVgOW/s0fdA5uzRXKS3LDsXWPkiQpakn11miw2AdJKO2i1+cUs0k0kQA+KqwILmhtIW0E7rrNzWURo3zigzgCg"
}
},
"unsigned": {
"age_ts": 1575738036831
}
}
|
|
For reference, the purge history docs are here.
Please note there are security advisories affecting this version, see https://github.com/matrix-org/synapse/security/advisories.
Can you find Synapse logs corresponding to this URL? I can see that synapse/synapse/storage/databases/main/events_worker.py Lines 507 to 514 in 21fea6b get_events_as_list returned an empty list. There are many possible explanations for this and it's hard to reason them through. Could you enable debug logging and try to reproduce the jump-to-timeline error for that event? It might give us more clues about what is going on here.
|
DMRobertson
changed the title
DMRobertson
added
A-Admin-API
A-Context-Endpoint
|
Closing due to lack of response. If you're the reporter please:
|
Description
After my db exploded in size (aprox 300gb) i needed to quickly get rid of as much state as possible so I purged every room from any non local event older than 30 days.
But now any room that has only local members is missing any message older than 30 days.
Looking into an example channel, I see 0 messages in element and a jump to last unread message button that results in
Now when I directly interact with postgres I can still find tons of events in this room with:
And also specifically event
$HvTnDTEDlze3Q15hditXICmEchw_ef7hehoKxx2O_Ukis still there.So far this seems to be true for all my local rooms, the events are still in the db, but not visible in Element. I can search for them on old clients, but I assume that's just because encrypted messages get cached for search.
Can I recover this somehow?
Steps to reproduce
I run my instance with workers, but during the purge all of those were disabled with only the main worker running and nginx also being disabled to avoid any other traffic to my instance.
I used this script to purge:
purge_script.zshHomeserver
dmnd.sh
Synapse Version
{"server_version":"1.71.0","python_version":"3.8.10"}
Installation Method
pip (from PyPI)
Database
postgres with a single server. i think i migrated from sqlite but that must have been 5 years ago.
Workers
Multiple workers
Platform
Configuration
No response
Relevant log output
Anything else that would be useful to know?
I don't have any logging from the purging itself since it happened too long ago and I haven't yet vacuumed so I am still running out of space. The log above is the only error I get repeatedly currently.
All of these rooms are room version 1-4 so very old
The text was updated successfully, but these errors were encountered: