Signed-off-by: Peter Somogyi <psomogyi@apache.org>
Signed-off-by: Xiaolin Ha <haxiaolin@apache.org>
(cherry picked from commit e555ac4a99419401e293385e78816f4c3f87e771)
When building against Hadoop 3.3.3 and any future version of Hadoop
incorporating reload4j the new Enforcer rule we have active in
branch-2.5 and up to exclude other logging frameworks besides log4j2
will trigger. We need to add exclusions to prevent that from
happening so the build will succeed.
Also exclude leveldbjni-all to avoid a LICENSE file generation error.
Add netty-all to hadoop-hdfs test context... to fix tests failing
trying to init minidfscluster.
Co-authored-by: stack <stack@apache.org>
Signed-off-by: Sean Busbey <busbey@apache.org>
Upon startup the shell logs a bunch of noise, mainly from zookeeper.
Also hide WARN level logging from the Hadoop libraries
Signed-off-by: Sean Busbey <busbey@apache.org>
Signed-off-by: Viraj Jasani <vjasani@apache.org>
This avoids starvation when the archive directory is large and takes a long time
to iterate through.
Signed-off-by: Duo Zhang <zhangduo@apache.org>
Signed-off-by: Anoop Sam John <anoopsamjohn@apache.org>
Signed-off-by: Pankaj <pankajkumar@apache.org>
HBASE-21812 replaced a for-loop with an each block.
Each block introduces a new scope, so a local variable defined inside it
cannot be accessed afterwards.
NameError: undefined local variable or method `admin' for main:Object
getServerNames at /opt/khp/hbase/bin/draining_servers.rb:81
addServers at /opt/khp/hbase/bin/draining_servers.rb:88
<main> at /opt/khp/hbase/bin/draining_servers.rb:146
This commit defines the admin local variable in the current scope beforehand,
so that we can still access it after the block.
Signed-off-by: Duo Zhang <zhangduo@apache.org>
(cherry picked from commit 2b573ae377be926922f48238c4da45a2289f318f)
Signed-off-by: Xiaolin Ha <haxiaolin@apache.org>
Signed-off-by: Nick Dimiduk <ndimiduk@apache.org>
(cherry picked from commit 487cf301a46a5df9201d24f19117b79967082b73)
This one adds a tracing context to the Master's ProcedureV2 execution. This ties all the spans
that result from a procedure execution back to a procedure by name. Maybe in a follow-on issue, we
can serialize span context along with procedures and display procedure start/stop/retry events and
parent-child hierarchy in tracing visualization.
Signed-off-by: Andrew Purtell <apurtell@apache.org>
Signed-off-by: Duo Zhang <zhangduo@apache.org>
Created OnheapDecodedCell and OffheapDecodedExtendedCell objects with duplicate copy of ByteBuffer's underlying array instead of original ByteBuffer
Signed-off-by: Andrew Purtell <apurtell@apache.org>
Signed-off-by: Pankaj Kumar<pankajkumar@apache.org>
(cherry picked from commit c198f23e5eb078abc47d4c3439477c6c2b4d5136)
TRACE level logging is expensive enough to warrant removal. They were
useful during development but are now just overhead.
Also we unnecessarily create new compressor and decompressor instances
in the reset() methods for the Aircompressor and Lz4 codecs. Remove.
Signed-off-by: Viraj Jasani <vjasani@apache.org>
Signed-off-by: Xiaolin Ha <haxiaolin@apache.org>
- introduce optional flag `hfile.pread.all.bytes.enabled` for pread that must read full bytes with the next block header
Signed-off-by: Josh Elser <elserj@apache.org>
There are two cases here:
1. Chaos Monkey thread died and there is no chaos after that.
2. Sometimes, regions are being moved back too quick that region server has not finished its initliazation yet.
wait sometime to make sure that region server finishes its initialization.
Signed-off-by: Wellington Chevreuil <wellington.chevreuil@gmail.com>
