Set Server.offline_since correctly

[garncarz]

I can see there's more than 500 servers stored in the production DB. Old ones are not getting removed and are contacted again and again during periodic tasks. That's probably the reason why Datadog metrics regarding DB refreshement are at so low levels (even 0) and the whole system gets overworked.

[garncarz]

✔️ Haha, this helped:

>>> from nogamespy import models
>>> models.remove_offline_entities()
[2018-11-21 15:48:17,920][DEBUG] nogamespy.models models.py:remove_offline_entities:168 | 3 offline players deleted.
[2018-11-21 15:48:17,923][DEBUG] nogamespy.models models.py:remove_offline_entities:171 | 518 servers went offline.
>>> models.remove_offline_entities()
>>> from nogamespy import settings
>>> settings.KEEP_OFFLINE_SERVERS_FOR_MINUTES = 0
>>> models.remove_offline_entities()
[2018-11-21 15:49:08,667][INFO] nogamespy.models models.py:remove_offline_entities:174 | 518 offline servers deleted.

[garncarz]

I needed to run the following, because there was a lot of periodic tasks waiting in the queue to be run. 😄 foreign_master_pulled metric jumped up to 39 and servers_refreshed even up to 97 (for a 20 minutes window).

$ docker-compose run redis redis-cli -h redis
redis:6379> flushdb
OK

[sonicpp]

Will be vietcong1.eu db refreshed more often, or is it another issue? Sometimes there are players who left 30 minutes ago...

[garncarz]

@sonicpp Yes, it should be. Because refreshing information for 500+ servers (with majority of them unaccessible, connection probably waiting for time-out... it's UDP) took some time. 🙂