[
https://issues.apache.org/jira/browse/HBASE-17938?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15985614#comment-15985614
]
Ted Yu commented on HBASE-17938:
--------------------------------
{code}
+ throw new IOException("Active session found, aborted command
execution");
{code}
Include information about the first session in message.
{code}
+ if (type == BackupType.FULL) {
+ deleteSnapshots(conn, backupInfo, conf);
+ cleanupExportSnapshotLog(conf);
+ }
+ restoreBackupTable(conn, conf);
+ deleteBackupTableSnapshot(conn, conf);
+ // clean up the uncompleted data at target directory if the ongoing
backup has already entered
+ // the copy phase
+ // For incremental backup, DistCp logs will be cleaned with the targetDir.
+ cleanupTargetDir(backupInfo, conf);
{code}
Assuming IOE may come out of each of the calls above, shouldn't a state machine
be designed for more robustness ?
Please put the next patch on review board.
> General fault - tolerance framework for backup/restore operations
> -----------------------------------------------------------------
>
> Key: HBASE-17938
> URL: https://issues.apache.org/jira/browse/HBASE-17938
> Project: HBase
> Issue Type: Sub-task
> Reporter: Vladimir Rodionov
> Assignee: Vladimir Rodionov
> Fix For: 2.0.0
>
> Attachments: HBASE-17938-v1.patch, HBASE-17938-v2.patch,
> HBASE-17938-v3.patch
>
>
> The framework must take care of all general types of failures during backup/
> restore and restore system to the original state in case of a failure.
> That won't solve all the possible issues but we have a separate JIRAs for
> them as a sub-tasks of HBASE-15277
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
