aasha commented on a change in pull request #949: HIVE-22990 Add file based ack
for replication
URL: https://github.com/apache/hive/pull/949#discussion_r392846451
##########
File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/ReplDumpTask.java
##########
@@ -126,21 +126,30 @@ public int execute() {
Path dumpRoot = new Path(conf.getVar(HiveConf.ConfVars.REPLDIR),
Base64.getEncoder().encodeToString(work.dbNameOrPattern.toLowerCase()
.getBytes(StandardCharsets.UTF_8.name())));
- Path currentDumpPath = new Path(dumpRoot, getNextDumpDir());
- Path hiveDumpRoot = new Path(currentDumpPath,
ReplUtils.REPL_HIVE_BASE_DIR);
- DumpMetaData dmd = new DumpMetaData(hiveDumpRoot, conf);
- // Initialize ReplChangeManager instance since we will require it to
encode file URI.
- ReplChangeManager.getInstance(conf);
- Path cmRoot = new Path(conf.getVar(HiveConf.ConfVars.REPLCMDIR));
- Long lastReplId;
- if (!dumpRoot.getFileSystem(conf).exists(dumpRoot)
- || dumpRoot.getFileSystem(conf).listStatus(dumpRoot).length ==
0) {
- lastReplId = bootStrapDump(hiveDumpRoot, dmd, cmRoot, hiveDb);
+ Path previousDumpMetaPath = getPreviousDumpMetadataPath(dumpRoot);
Review comment:
yes only hive user should have access to this staging directory as we are
completely relying on FS based ack.
But in case of irrecoverable users, users should delete the _load_execption
file for dump/load to proceed. So they will need some admin access. But in
general they shouldn't modify anything.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
