Closed samijuhani closed 4 years ago
These got stuck in the cache for some reason, which resulted in more not loading in. Cache cleared now.
Still missing quite alot
I'm deploying a fix now that will cache cancellations less, so these should show up soon in all enviuronments (live in dev already).
Might need @hellej for this as seems mentioned alerts above are missing from database
So looks like we are missing cancelled journeys that have been cancelled after the end time of the journey and partly cancelled journeys.
So looks like we are missing cancelled journeys that have been cancelled after the end time of the journey and partly cancelled journeys.
@samijuhani @danieldunderfelt Is this happening in both dev and prod or only prod?
Currently, more cancellations are saved to dev as also cancellations added to the past are queried from OMM (by HSLdevcom/transitdata-omm-cancellation-source/pull/16). That feature/fix was not yet merged to master since there was a concern (some time ago) that some cancellations may be missing from dev because of it (@samijuhani ?). However, I can confirm that more cancellations have been saved to dev than prod e.g. last_modified > '2019-12-05 00:00:00.000+00' and last_modified < '2019-12-09 00:00:00.000+00'
-> 41 in prod vs. 46 in dev.
I can deploy the new cancellation source to stage & prod tomorrow. Will close #486 too
Cancelled journeys after journey time has been fixed already in prod also. Only thing to do now with cancelled are partial cancellations
Except https://github.com/HSLdevcom/transitlog-ui/issues/486 seems still not to be there in PROD (nor DEV).
For example this has been cancelled afterwards and is seen in every env https://reittiloki.hsl.fi/journey/20191211/064500/1092/2/0022_881?date=2019-12-11&mapView=60.21502765949603%2C25.076079368591312&route.routeId=1092&route.direction=2&route.originStopId=1454128&tab=&details-tab=journey-alerts&mapZoom=15&vehicle=0022%2F881&live=false&time=06%3A45%3A00
The cancellation in #468 is still missing from envs where it doesn't exist since cancellations (and all events) are only sent once. So if the environment doens't catch it, there's no mechanism to resend it except republishing it or manually entering it into the database. Am I correct, @hellej?
@danieldunderfelt @samijuhani What we know about the cancellation mentioned in #486:
Example today: