[HBASE-25549] Provide a switch that allows avoiding reopening all regions when modifying a table to prevent RIT storms.

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.0.0-alpha-1
• Fix Version/s: 2.6.0, 3.0.0-beta-1, 2.5.7
• Component/s: master, shell

Release Note

New APIs are added to Admin, AsyncAdmin, and hbase shell to allow modifying a table without reopening all regions. Care should be used in using this API, as regions will be in an inconsistent state until they are all reopened. Whether this matters depends on the change, and some changes are disallowed (such as enabling region replication or adding/removing a column family).

• Assignee: Zhuoyue Huang
• Reporter: Zhuoyue Huang
• Created: 04/Feb/21 04:06
• Updated: 03/Dec/23 04:43
• Resolved: 01/Dec/23 20:10

Description

Under normal circumstances, modifying a table will cause all regions belonging to the table to enter RIT. Imagine the following two scenarios：

1. Someone entered the wrong configuration （e.g. negative 'hbase.busy.wait.multiplier.max' value） when altering the table, causing thousands of online regions to fail to open, leading to online accidents.
2. Modify the configuration of a table, but this modification is not urgent, the regions are not expected to enter RIT immediately.

'alter_lazy' is a new command to modify a table without reopening any online regions except those regions were assigned by other threads or split etc.

 

Provide an optional lazy_mode for the alter command to modify the TableDescriptor without the region entering the RIT. The modification will take effect when the region is reopened.

Issue Links

• links to: GitHub Pull Request #2924
• links to: GitHub Pull Request #5549

Sub-Tasks

• Provide the switch to avoid reopening regions in the alter sync command: Resolved
• Port the switch to avoid reopening regions in the alter async in HBase 2: Resolved