Sign InTry Free

Troubleshoot Log Backup

This document summarizes common problems during log backup and the solutions.

After restoring a downstream cluster using the br restore point command, data cannot be accessed from TiFlash. What should I do?

In v6.2.0, PITR does not support restoring the TiFlash replicas of a cluster. After restoring data, you need to execute the following statement to set the TiFlash replica of the schema or table.

ALTER TABLE table_name SET TIFLASH REPLICA count;

What should I do if the status of a log backup task becomes ERROR?

During a log backup task, the task status becomes ERROR if it fails and cannot be recovered after retrying. The following is an example:

br log status --pd x.x.x.x:2379

● Total 1 Tasks.
> #1 <
                    name: task1
                  status: ○ ERROR
                   start: 2022-07-25 13:49:02.868 +0000
                     end: 2090-11-18 14:07:45.624 +0000
                 storage: s3://tmp/br-log-backup0ef49055-5198-4be3-beab-d382a2189efb/Log
             speed(est.): 0.00 ops/s
      checkpoint[global]: 2022-07-25 14:46:50.118 +0000; gap=11h31m29s
          error[store=1]: KV:LogBackup:RaftReq
error-happen-at[store=1]: 2022-07-25 14:54:44.467 +0000; gap=11h23m35s
  error-message[store=1]: retry time exceeds: and error failed to get initial snapshot: failed to get the snapshot (region_id = 94812): Error during requesting raftstore: message: "read index not ready, reason can not read index due to merge, region 94812" read_index_not_ready { reason: "can not read index due to merge" region_id: 94812 }: failed to get initial snapshot: failed to get the snapshot (region_id = 94812): Error during requesting raftstore: message: "read index not ready, reason can not read index due to merge, region 94812" read_index_not_ready { reason: "can not read index due to merge" region_id: 94812 }: failed to get initial snapshot: failed to get the snapshot (region_id = 94812): Error during requesting raftstore: message: "read index not ready, reason can not read index due to merge, region 94812" read_index_not_ready { reason: "can not read index due to merge" region_id: 94812 }

To address this problem, check the error message for the cause and perform as instructed. After the problem is addressed, run the following command to resume the task:

br log resume --task-name=task1 --pd x.x.x.x:2379

After the backup task is resumed, you can check the status using br log status. The backup task continues when the task status becomes NORMAL.

● Total 1 Tasks.
> #1 <
              name: task1
            status: ● NORMAL
             start: 2022-07-25 13:49:02.868 +0000
               end: 2090-11-18 14:07:45.624 +0000
           storage: s3://tmp/br-log-backup0ef49055-5198-4be3-beab-d382a2189efb/Log
       speed(est.): 15509.75 ops/s
checkpoint[global]: 2022-07-25 14:46:50.118 +0000; gap=6m28s

What should I do if the error message ErrBackupGCSafepointExceeded is returned when using the br log resume command to resume the suspended task?

Error: failed to check gc safePoint, checkpoint ts 433177834291200000: GC safepoint 433193092308795392 exceed TS 433177834291200000: [BR:Backup:ErrBackupGCSafepointExceeded]backup GC safepoint exceeded

After you pause a log backup task, to prevent the MVCC data from being garbage collected, the pausing task program sets the current checkpoint as the service safepoint automatically. This ensures that the MVCC data generated within 24 hours can remain. If the MVCC data of the backup checkpoint has been generated for more than 24 hours, the data of the checkpoint will be garbage collected, and the backup task is unable to resume.

To address this problem, delete the current task using br log stop, and then create a log backup task using br log start. At the same time, you can perform a full backup for subsequent PITR.

What should I do if an error occurs when executing the Exchange Partition DDL during PITR log restoration?

The following error occurs during PITR log restoration:

restore of ddl `exchange-table-partition` is not supported

In v6.2.0, the log backup feature is not compatible with the Exchange Partition DDL. It is recommended not to use this DDL during log backup. If you have executed the DDL, you must perform a full backup immediately. Then, PITR can restore the log data after the full backup checkpoint.

Download PDFRequest docs changes
Was this page helpful?
Open Source Ecosystem
TiDB
TiKV
TiSpark
Chaos Mesh
© 2022 PingCAP. All Rights Reserved.