[HBASE-21365] Throw exception when user put data with skip wal to a table which may be replicated - ASF JIRA

Voters

Watch issue

Watchers

Create sub-task

Link

Clone

Update Comment Author

Replace String in Comment

Update Comment Visibility

Delete Comments

XML

Word

Printable

JSON

Details

Type: Improvement
Status: Resolved
Priority: Minor
Resolution: Fixed
Affects Version/s: 3.0.0-alpha-1
Fix Version/s: 3.0.0-alpha-1
Component/s: Client
Labels:
None

Hadoop Flags:

Incompatible change
Release Note:

Hide
If a table's column family's replication scope is not REPLICATION_SCOPE_LOCAL, this table may be replicated to peer cluster. As hbase's replication rely on WAL, if put data with Durability.SKIP_WAL to this table, the regionserver will throw a new InvalidMutationDurabilityException to client.

Show
If a table's column family's replication scope is not REPLICATION_SCOPE_LOCAL, this table may be replicated to peer cluster. As hbase's replication rely on WAL, if put data with Durability.SKIP_WAL to this table, the regionserver will throw a new InvalidMutationDurabilityException to client.

Description

A real problem in our production cluster. A user point that his table's data can't be replicate to the peer cluster. Then we start to debug the reason. We checked the replication scope, checked the replication wal entry filter, and check the namespace,tablecfs config. But didn't found any problem. We enabled the RS's debug log to find the reason. Finally, we found use use put with skip wal to write data. But it taked a long time... Our replication use wal to replicate data. So the data can't be replicated to peer cluster. I thought throw a exception may be better for user if the table's replication scope is not 0. (as 0 means not replicated).