Hello Percona folks! Brian Matheson from MayaData here. I have some general questions about Postgres on k8s as well as some questions specific to Percona’s offerings.
First off, I’m running under k8s with a 5.4 kernel using a recent helm chart (in my case Bitnami’s but open to suggestions :^). When I first deployed the db server went into a crash loop. I checked things out and found a debug option for the chart that purported to enable more logging, and indeed I got a bus error message (weird right?).
So, after a little googling I came across a bus error in another instance where postgres was running with huge_pages enabled. Aha! Super simple right, just drop that huge_pages=off into the bottom of the config and we’re golden. Only I’m still getting the bus error :^/ Assuming it’s even related (which is an assumption at this point) I thought maybe fixing permissions on the huge_pages sys interface might be a good direction to go in.
Do you have a chart or operator that does the right thing regarding huge pages?
Is there an open-source helm chart from Percona? Or a trial of the commercial version?
Thanks in advance!
Brian
Hi Brian,
As of now, we don’t have support for k8s for PostgreSQL (helm charts or operator). But it may come down the like as we are considering Kubernetes as one of the important platforms.
Regarding the huge_pages, it is better to have ‘try’ option which is the default for PostgreSQL. With huge_pages set to try, the server will try to request huge pages, but fall back to the default if that fails.
PostgreSQL: Documentation: 14: 20.4. Resource Consumption
The huge pages generally found to make a difference when we have close to 100GB RAM and once the page table is big.Regards,
Jobin.