Is there a doc on how fleet uses MySQL in terms of creating and reusing connections and other such lifecycles. We are trying to scale fleet to ~35K osquery hosts and any guidance on how to scale the DB to handle this, specifically what kind of access patterns fleet has / makes to the DB would be very valuable. We are hosting the MySQL (mariadb) on-prem so the guide given in the docs regarding AWS kind of doesn't exactly apply to us

@Smaran Jawalkar mariadb is not a an officially supported deployment path from Fleets standpoint so unfortunately we don’t have a ton of subject matter expertise here. we actually frequently hear about issues and incompatibilities on that front 😞 the aws reference architecture provided is our best practice recommendation. however maybe other community members have a different experience and can chime in…

That is sad to hear. Does fleet officially support any other on-prem hostable DB? Or is it just MySQL (

["mysql:8.0.36", "mysql:8.4.3", "mysql:9.1.0"]

) as mentioned int he docs, apart from aurora

Just MySQL

Hi @Smaran Jawalkar! We don't have any specific documentation about access patterns. Generally, I'd recommend looking at the specs for the instances outlined in the reference architecture and using that as a baseline. For 35k hosts, The database is used fairly intensely, so I'd recommend starting with a MySQL cluster with two nodes with one CPU and 32GB of RAM available for each. From there, you can tweak things as needed. You may be able to scale the RAM down a bit if you find that you aren't pushing the memory too hard.

Tysm! @Kathy Satterlee

Any time :)