As the webrequest dataset source has been migrated to haproxykafka (on 1st of April), we're now ready to switch off varnishkafka instances that are still producing and sending data to the "old" kafka topics.
If everyone involved agrees, I would define the following roadmap:
- On Monday (12 May 2025) I'll disable and stop the varnishkafka service on both text and upload cluster in MAGRU DC, to test eventual and unforeseen issues with monitoring, configuration or performances.
- On the next week (19-23 May 2025) we can switch off varnishkafka instances on all cache hosts, on all datacenters. This is a good time to check that everything on the DE side works as expected and (eventually) do some fixes.
- On June the 19th we will remove entirely varnishkafka, and all related configurations, from production hosts.
Let me know if this sounds like a feasible roadmap for all involved people
- Shutdown varnishkafka (webrequest) in MAGRU DC
- Shutdown varnishkafka (webrequest) instances on all cp hosts
- Remove varnishkafka (webrequest) instances and related configuration from all cp hosts