Stream loading hangs on whole pod

Hi
viewing activities, mentions and other users profile works fine.
but the stream view simply hangs in some timeouts.

as per this suggestion: https://github.com/diaspora/diaspora/issues/8051 already disabled the chat.

but this does not seem to bring the stream view back again.

any help is appreciated.

thx
hk

If you have disabled the chat and the stream is not loading, you will not see a different error message in the JS console. Please share that.

thx, we are now here:

content.js:4 [Deprecation] chrome.loadTimes() is deprecated, instead use standardized API: nextHopProtocol in Navigation Timing 2. https://www.chromestatus.com/features/5637885046816768.
(anonymous) @ content.js:4
content.js:5 [Deprecation] chrome.loadTimes() is deprecated, instead use standardized API: nextHopProtocol in Navigation Timing 2. https://www.chromestatus.com/features/5637885046816768.
(anonymous) @ content.js:5
deadsuperhero.com/photo/profile/50/5.jpg:1 Failed to load resource: net::ERR_CERT_COMMON_NAME_INVALID
therealtalk.s3.amazonaws.com/uploads/images/thumb_small_7231bcf6dded5a6a327f.jpeg:1 Failed to load resource: the server responded with a status of 404 (Not Found)
stream:1 Failed to load resource: the server responded with a status of 504 (Gateway Time-out)

seems like a blatant timeout here.

Given that https://diaspora.kapper.net/u/hknet loads just fine, this is probably a simple load-issue. Is your database-server excessively busy or is there lots of disk I/O? The stream queries are the most expensive queries diaspora can run, so it’s no surprise those timeout first.

well, it’s a small but dedicated box and at iotop we see up to 50MB/s reads for postgres but usually we are at 1-5 MB/s reads it seems this started (by user reports) with the latest update to 0.7.12.0 but holidays had me see this issue reports only and I have no real idea except moving this box to SSDs :slight_smile:

well, getting desperate I did a reindex of the whole database (had to shutdown diasp* for that of course); then restarted diasp and first got a:
2019-08-28 17:39:35 CEST [21733-1] diaspora@diaspora_production LOG: could not send data to client: Broken pipe
2019-08-28 17:39:35 CEST [21733-2] diaspora@diaspora_production FATAL: connection to client lost
Waiting for data… (interrupt to abort

but after some more waiting and trying the stream is loading again, though I guess we have to move this pod to some bigger iron soon as load time increases quite a bit.

Which version of Postgres are you using? Switching from 9.6 to 11 solved most of my performance problems on Framasphere. And 12 just came out yesterday (but I didn’t test it yet).