Major compaction timeout
The major compaction timeout duration is specified by zone_merge_timeout. The default value is 3 hours.
When a major compaction times out, the major compaction enters the TIMEOUT state. You can query the __all_zone table for the major compaction status. When a major compaction times out, the compaction continues but is tagged with a timeout flag bit. The purpose of the timeout flag bit is to notify the user that the compaction will last for a period longer than expected and will continue in the peak hours in a business day. It indicates that a large number of data changes were made in the day before the major compaction, for example, a large amount of data was imported.
When a major compaction times out, you do not need to take additional measures, and the major compaction continues. The administrator can check the progress of the major compaction, the compression algorithm of the table, and the data changes in the previous day. If a large number of abnormal data changes were made, you can wait for the compaction to complete. If major compaction timeouts occur frequently when the data volume is normal, you can modify the major compaction parameters to speed up the compaction. For example, you can increase the number of threads for a major compaction.
Merge errors
When an exception occurs during the major compaction process, the major compaction enters the ERROR state. The major compaction stops when a compaction error occurs. When a compaction error occurs, you need to contact the administrator to rectify the error. A major compaction can be re-initiated after the error is rectified. Common compaction errors include a full disk, I/O timeout, disk I/O error, and compaction canceled.