Sign InTry Free

Perform a Rolling Update to a TiDB Cluster in Kubernetes

When you perform a rolling update to a TiDB cluster in Kubernetes, the Pod is shut down and recreated with the new image or/and configuration serially in the order of PD, TiKV, TiDB. Under the highly available deployment topology (minimum requirements: PD * 3, TiKV * 3, TiDB * 2), performing a rolling update to PD and TiKV servers does not impact the running clients.

  • For the clients that can retry stale connections, performing a rolling update to TiDB servers neither impacts the running clients.
  • For the clients that can not retry stale connections, performing a rolling update to TiDB servers will close the client connections and cause the request to fail. For this situation, it is recommended to add a function for the clients to retry, or to perform a rolling update to TiDB servers in idle time.

Upgrade using TidbCluster CR

If the TiDB cluster is deployed directly using TidbCluster CR, or deployed using Helm but switched to management by TidbCluster CR, you can upgrade the TiDB cluster by the following steps.

Upgrade the version of TiDB using TidbCluster CR

  1. Modify the image configurations of all components in TidbCluster CR.

    Usually, components in a cluster are in the same version. You can upgrade the TiDB cluster simply by modifying spec.version. If you need to use different versions for different components, configure spec.<pd/tidb/tikv/pump/tiflash/ticdc>.version.

    The version field has following formats:

    • spec.version: the format is imageTag, such as v5.0.6

    • spec.<pd/tidb/tikv/pump/tiflash/ticdc>.version: the format is imageTag, such as v3.1.0

    kubectl edit tc ${cluster_name} -n ${namespace}
  2. Check the upgrade progress:

    watch kubectl -n ${namespace} get pod -o wide

    After all the Pods finish rebuilding and become Running, the upgrade is completed.

Force an upgrade of TiDB cluster using TidbCluster CR

If the PD cluster is unavailable due to factors such as PD configuration error, PD image tag error and NodeAffinity, then scaling the TiDB cluster, upgrading the TiDB cluster and changing the TiDB cluster configuration cannot be done successfully.

In this case, you can use force-upgrade to force an upgrade of the cluster to recover cluster functionality.

First, set annotation for the cluster:

kubectl annotate --overwrite tc ${cluster_name} -n ${namespace} tidb.pingcap.com/force-upgrade=true

Change the related PD configuration to make sure that PD is in a normal state.

Modify TiDB cluster configuration

  1. Refer to Configure TiDB components to modify the component configuration in the TidbCluster CR of the cluster:

    kubectl edit tc ${cluster_name} -n ${namespace}
  2. View the updating progress after the configuration is modified:

    watch kubectl -n ${namespace} get pod -o wide

    After all the Pods are recreated and are in the Running state, the configuration is successfully modified.

Download PDF
Playground
New
One-stop & interactive experience of TiDB's capabilities WITHOUT registration.
Products
TiDB
TiDB Dedicated
TiDB Serverless
Pricing
Get Demo
Get Started
© 2024 PingCAP. All Rights Reserved.
Privacy Policy.