saintstack commented on a change in pull request #1955:
URL: https://github.com/apache/hbase/pull/1955#discussion_r444302316
##########
File path:
hbase-server/src/main/java/org/apache/hadoop/hbase/wal/BoundedRecoveredHFilesOutputSink.java
##########
@@ -191,50 +186,22 @@ public boolean keepRegionEvent(Entry entry) {
return false;
}
+ /**
+ * @return Returns a base HFile without compressions or encodings; good
enough for recovery
Review comment:
Asking the Master for the table descriptor first and if it fails, read
the fs is what we had before this patch. Problem was that default configuration
had it that the RPC query wasn't timing out before the below happened:
2020-06-18 19:53:54,175 ERROR [main] master.HMasterCommandLine: Master
exiting
java.lang.RuntimeException: Master not initialized after 200000ms
I could dick around w/ timings inside here so RPC timed out sooner. It uses
the server's Connection. I suppose I could make a new Connection everytime we
want to query a Table schema and configure it to try once only but that is a
lot of work to do at recovery time.
On the compactions, don't we usually pick up the small files first?
----------------------------------------------------------------
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]
