Backend cache server connection pooling

Wednesday, 13 November 13:35 UTC
Changelog DevOps

We saw an unusual bug yesterday evening where one customer status page was not correctly flushed from our CDN cache when a notice was updated.

Whilst it has been difficult to pin down the exact cause it was noted that the way we manage connections to our backend cache server could be improved.

This afternoon we have rolled out a small patch that improves this, ensuring all our backend processes now use a shared connection pool to the cache servers, making them more reliable.

Avatar for Robert Rawlins
Robert Rawlins