POI API unavailable?

We’ve been developing an app using the API for the past two weeks, demonstration is tomorrow and now the API is suddenly returning a timeout error 524. Nothing has changed at our end. Is the API server down ?

1 Like

Hi, yes - our API has a couple of data store layers split into a caching layer and an sql database layer.

The SQL database server experienced issues which eventually affected our ability to update the cache. As our (cache) Mongo DB instance has been known to fall over we have logic that says when our cache gets old (out of date by several hours) we then revert back to querying the sql database directly, unfortunately this assumes that our sql database is basically more reliable that our cache database, which wasn’t true in this case.

Various servers are being resized and restarted, the cache is being rebuilt now.

1 Like

Thank you Christopher. Will the server be running normally again before 5PM GMT today ?

Yes I expect so, the database server is being restarted now - it’s still running oddly but if we can get the cache to fully rebuild from the database then the database is no longer involved.

Note that the Open Charge Map API is not suitable for heavy production use (but some users ignore that and make very heavy use of the API, impacting other users) and does not come with an SLA, we recommend importing the data into your own database for any intensive applications (which your application may not be).

1 Like

Thanks for the advice. The project is a technology demonstrator using only front-end JS so we don’t have the option of ingesting the database or even using Node.JS

Sure thing, the database has returned to normal now and the cache is being rebuilt. We still have an issue where not all items in the DB are being cached but if your demo doesn’t rely on specific POIs then it will be fine.

Thank you { breathes again }

Hi !
Do you know when the situation will return to normal ?
There is always missing charging stations around me.

Thanks :ok_hand:

Hi, this issue is now resolved. The problem was caused by a user submitting a large (greater than 16MB) Russian spam comment in one of the submissions, which then affected the rebuild of the cache. We have added extra validation to prevent that.