Troubleshooting TiUP

This article introduces some common faults and troubleshooting methods during the use of TiUP. If this article does not cover the problems you are currently encountering, you can ask for help in the following ways:

1. Github Issues Create a new issue.
2. Submit your question at AskTUG.

1. TiUP command troubleshooting

1.1 use tiup list can't see the latest component list

TiUP does not update the latest component list from the mirror server every time. You can force a refresh of the component list through tiup list.

1.2 use tiup list can not see the latest version information of a component

Same as 1.1, the version information of the component will only be obtained from the mirror server if there is no local cache. You can refresh the component list through tiup list .

1.3 Interrupted in the process of downloading components

If the network is interrupted during the download of the component, it may be caused by the unstable network. You can try to download it again. If the download fails for many times, please feedback to Github Issues, possibly due to CDN server.

1.4 Checksum error occurred during component download

Because the CDN will have a short cache time, the new checksum file does not match the component package. It is recommended to try again after 5 minutes. If it still does not match, please feedback to Github Issues.

2. TiUP Cluster component troubleshooting

2.1 Prompt during deployment unable to authenticate, attempted methods [none publickey]

Due to uploading the component package to the remote host and initializing it during deployment, this process needs to connect to the remote host. This error is caused by the SSH private key that is not found to connect to the remote host.Make sure you specify the private key via tiup cluster deploy -i identity_file .

1. If the -i parameter is not specified, it may be because TiUP does not automatically find the private key path. It is recommended to explicitly specify the private key path through -i.
2. If the -i parameter is specified, it may be that the specified private key cannot be logged in, which can be verified by manually executing the ssh -i identity_file [email protected] command.
3. If you log in to the remote host with a password, make sure to specify the -p parameter and enter the correct login password.

2.2 Upgrading interruption with TiUP Cluster

In order to avoid user misuse, TiUP Cluster does not support the upgrade of specified nodes. Therefore, after the upgrade fails, the upgrade operation needs to be performed again, including the idempotent operation during the upgrade process.

The upgrade operation will be divided into the following steps:

1. Back up the old version components of all nodes first
2. Distribute new components to remote
3. Rolling restart of all components

If the upgrade operation is interrupted during rolling restart, instead of repeating the tiup cluster upgrade operation, you can use tiup cluster restart -N -N to restart the node that did not complete the restart.If the number of unrestarted nodes of the same component is relatively large, you can also restart a certain type of component through tiup cluster restart -R .

2.3 upgrade discovery node_exporter-9100.service/blackbox_exporter-9115.service does not exist

This situation may be caused by the fact that the previous cluster was migrated by TiDB Ansible, and TiDB Ansible did not deploy the exporter before.To work around this situation, you can temporarily copy the missing files from the other node to the new node manually.Later, we will complete the missing components during the migration process.