Define environment variables¶

Sometimes you need to define new environment variables to provide additional configuration for the components of your cluster. For example, you can use it to customize the configuration of HAProxy, or to add additional options for PMM Client.

The Operator can store environment variables in Kubernetes Secrets . Here is an example with several options related to HAProxy:

apiVersion: v1
kind: Secret
metadata:
  name: my-env-var-secrets
type: Opaque
data:
  HA_CONNECTION_TIMEOUT: MTAwMA==
  OK_IF_DONOR: MQ==
  HA_SERVER_OPTIONS: Y2hlY2sgaW50ZXIgMzAwMDAgcmlzZSAxIGZhbGwgNSB3ZWlnaHQgMQ==

Note

Variables used in this example have the following effect:

HA_CONNECTION_TIMEOUT allows to set custom timeout for health checks done by HAProxy (it repeatedly executes a simple status query on XtraDB Cluster instances). The default 10 seconds timeout is good for most workloads, but increase should be helpful in case of unstable Kubernetes network or soft lockups happening on Kubernetes nodes.

OK_IF_DONOR allows application connections to XtraDB Cluster donors. The backup is running on the donor node, and SQL queries combined with it could run slower than usual. Enable the option to grant application access when there is only one XtraDB Cluster node alive, and a second XtraDB Cluster node is joining the cluster via SST.

HA_SERVER_OPTIONS allows to set the custom options for the server in the HAProxy configuration file. You can start with the default check inter 30000 rise 1 fall 5 weight 1 set, and add required options referenced in the upstream documentation .

As you can see, environment variables are stored as data - i.e., base64-encoded strings, so you’ll need to encode the value of each variable. For example, To have HA_CONNECTION_TIMEOUT variable equal to 1000, you can run echo -n "1000" | base64 --wrap=0 (or just echo -n "1000" | base64 in case of Apple macOS) in your local shell and get MTAwMA==.

Note

Similarly, you can read the list of options from a Base64-encoded string:

$ echo "MTAwMA==" | base64 --decode

When ready, apply the YAML file with the following command:

$ kubectl create -f deploy/my-env-secret.yaml

Put the name of this Secret to the envVarsSecret key either in pxc, haproxy or proxysql section of the deploy/cr.yaml` configuration file:

haproxy:
  ....
  envVarsSecret: my-env-var-secrets
  ....

Now apply the deploy/cr.yaml file with the following command:

$ kubectl apply -f deploy/cr.yaml

Another example shows how to pass LD_PRELOAD environment variable with the alternative memory allocator library name to mysqld. It’s often a recommended practice to try using an alternative allocator library for mysqld in case the memory usage is suspected to be higher than expected, and you can use jemalloc allocator already present in Percona XtraDB Cluster Pods with the following environment variable:

LD_PRELOAD=/usr/lib64/libjemalloc.so.1

Create a new YAML file with the contents similar to the previous example, but with LD_PRELOAD variable, stored as base64-encoded strings:

apiVersion: v1
kind: Secret
metadata:
  name: my-new-env-var-secrets
type: Opaque
data:
  LD_PRELOAD: L3Vzci9saWI2NC9saWJqZW1hbGxvYy5zby4x

If this YAML file was named deploy/my-new-env-var-secret, the command to apply it will be the following one:

$ kubectl create -f deploy/my-new-env-secret.yaml

Now put the name of this new Secret to the envVarsSecret key in pxc section of the deploy/cr.yaml` configuration file:

pxc:
  ....
  envVarsSecret: my-new-env-var-secrets
  ....

Don’t forget to apply the deploy/cr.yaml file, as usual:

$ kubectl apply -f deploy/cr.yaml

Last update: 2025-04-15