Could you share your config file Juergen? I have a keepalive but in the haproxies.
Kind regards On Mon, Sep 16, 2024 at 3:03 PM Jürgen Gotteswinter <juergen.gotteswin...@mgm-tp.com.invalid> wrote: > Hello, > > i am currently also working on such a setup, like you said it causes > issues when distributing the write requests. My solution is to use > keepalived on each cluster node to make sure writes only go to a single > node (VIP with NAT 3306 -> 127.0.0.1:3306). > > So far, no problems. > > Cheers, > > Juergen > > Am 13.09.24, 22:51 schrieb "Francisco Arencibia Quesada" < > arencibia.franci...@gmail.com <mailto:arencibia.franci...@gmail.com>>: > > > Hello Guys, > > > I am currently using Percona XtraDB Cluster for my database, but CloudStack > relies on functions like GET_LOCK and RELEASE_LOCK, which are not > recommended in a Galera-based cluster. Setting pxc_strict_mode = PERMISSIVE > allows CloudStack to function, but this causes issues with replication > consistency and stability. If I set pxc_strict_mode = ENFORCING, CloudStack > stops working because it uses these functions, but setting it to PERMISSIVE > breaks the replication in the cluster. > > > Is there any way to configure Percona XtraDB Cluster to allow CloudStack to > work properly without compromising the integrity of the replication? > > > I use CS 4.18, 4.19 both with the same problem. > Thanks a lot in advance > :) > -- > *Francisco Arencibia Quesada.* > *DevOps Engineer* > > > > -- *Francisco Arencibia Quesada.* *DevOps Engineer*
