Add a check for hbase:meta being online before we go to read it.
If not online, move into a holding-pattern until rectified, probably
by external operator.
Incorporates bulk of patch made by Allan Yang over on HBASE-21035.
M hbase-common/src/main/java/org/apache/hadoop/hbase/util/RetryCounterFactory.java
Add a Constructor for case where retries are for ever.
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/HMaster.java
Move stuff around so that the first hbase:meta read is the AM#loadMeta.
Previously, checking table state and/or favored nodes could end up
trying to read a meta that was not onlined holding up master startup.
Do similar for the namespace table. Adds new methods isMeta and
isNamespace which check that the regions/tables are online.. if not,
we wait logging with a back-off that assigns need to be run.
Signed-off-by: Allan Yang <allan163@apache.org>
Signed-off-by: Duo Zhang <zhangduo@apache.org>
HBASE-21113 Apply the branch-2 version of HBASE-21095, The timeout retry
logic for several procedures are broken after master restarts
I applied the patch HBASE-21095 and then reverted it so could apply the
patch as HBASE-21113 (by reverting the HBASE-21095 revert but pushing
with this message!).
This reverts commit 4978db8102.
Write the hbase-1.x hbck1 lock file to block out hbck1 instances writing
state to an hbase-2.x cluster (could do damage).
Set hbase.write.hbck1.lock.file to false to disable this writing.
Look for the particular case where RS does the close of region w/o
involving Master and log special message in this case. Dodgy. But
until we have Master run shutdown of all regions, better than
the message we currently show.
With this change if hbase.wal.meta_provider is not explicitly set,
it uses whatever set with hbase.wal.provider. this change avoids a use
case of unexpectedly using two different providers when only
hbase.wal.provider is set to non-default but not hbase.wal.meta_provider.
This change also include document (architecture.adoc) update
Also, this is a port from master to branch-2
Signed-off-by: Zach York <zyork@apache.org>
Signed-off-by: Michael Stack <stack@apache.org>
Signed-off-by: Sean Busbey <busbey@apache.org>
Signed-off-by: Duo Zhang <Apache9@apache.org>
This patch switches to the builder pattern by adding a helper method.
It also checks to ensure that the pattern is available (i.e. that
HBase is running on a hadoop version that supports it).
Amending-Author: Mike Drob <mdrob@apache.org>
Signed-off-by: tedyu <yuzhihong@gmail.com>
Signed-off-by: zhangduo <zhangduo@apache.org>
Make the #copyCellInto method smaller so it inlines; we do it by
checking for the common type early and then taking a code path
that presumes ByteBufferExtendedCell -- avoids checks.
- -jar parameter now accepts multiple jar files and directories of jar files.
- observer classes can be verified by -class option.
- -table parameter was added to check table level coprocessors.
- -config parameter was added to obtain the coprocessor classes from
HBase cofiguration.
- -scan option was removed.
Signed-off-by: Mike Drob <mdrob@apache.org>
With prefetch-on-open enabled, the task doing the prefetching was using
non-positional (i.e. streaming) reads. If the main (non-prefetch) thread
was also using non-positional reads, these two would conflict, because
inputstreams are not thread-safe for non-positional reads.
In the case of an encrypted filesystem, this could cause JVM crashes,
etc, as underlying cipher buffers were freed underneath the racing
threads. In the case of a non-encrypted filesystem, less severe errors
would be thrown. The included unit test reproduces the latter case.
(cherry picked from commit 025ddce868)
Signed-off-by: Todd Lipcon <todd@cloudera.com>
Also changes modify table operations to help the case where a MTP spans
two master, avoiding the sanity-checks propagating back to the client
unnecessarily.
Signed-off-by: Josh Elser <elserj@apache.org>
Signed-off-by: Michael Stack <stack@apache.org>
ModifyTableProcedure is using MoveRegionProcedure in a way
that was unintended from the original implementation. As such,
we have to guard against certain usages of it. We know we can
re-open OPEN regions, but regions in OPENING will similarly
soon be OPEN (thus, we want to reopen those regions too).
Signed-off-by: Michael Stack <stack@apache.org>
Signed-off-by: zhangduo <zhangduo@apache.org>
* modify the jar checking script to take args; make hadoop stuff optional
* separate out checking the artifacts that have hadoop vs those that don't.
* * Unfortunately means we need two modules for checking things
* * put in a safety check that the support script for checking jar contents is maintained in both modules
* * have to carve out an exception for o.a.hadoop.metrics2. :(
* fix duplicated class warning
* clean up dependencies in hbase-server and some modules that depend on it.
* allow Hadoop to have its own htrace where it needs it
* add a precommit check to make sure we're not using old htrace imports
Conflicts:
hbase-backup/pom.xml
hbase-checkstyle/src/main/resources/hbase/checkstyle-suppressions.xml
Signed-off-by: Mike Drob <mdrob@apache.org>
Cannot go to latest (8.9) yet due to
https://github.com/checkstyle/checkstyle/issues/5279
* move hbaseanti import checks to checkstyle
* implment a few missing equals checks, and ignore one
* fix lots of javadoc errors
Signed-off-by: Sean Busbey <busbey@apache.org>
A reattempt at fixing HBASE-20173 [AMv2] DisableTableProcedure concurrent to ServerCrashProcedure can deadlock
The scenario is a SCP after processing WALs, goes to assign regions that
were on the crashed server but a concurrent Procedure gets in there
first and tries to unassign a region that was on the crashed server
(could be part of a move procedure or a disable table, etc.). The
unassign happens to run AFTER SCP has released all RPCs that
were going against the crashed server. The unassign fails because the
server is crashed. The unassign used to suspend itself only it would
never be woken up because the server it was going against had already
been processed. Worse, the SCP could not make progress because the
unassign was suspended with the lock on a region that it wanted to
assign held making it so it could make no progress.
In here, we add to the unassign recognition of the state where it is
running post SCP cleanup of RPCs. If present, unassign moves to finish
instead of suspending itself.
Includes a nice unit test made by Duo Zhang that reproduces nicely the
hung scenario.
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/FailedRemoteDispatchException.java
Moved this class back to hbase-procedure where it belongs.
M hbase-procedure/src/main/java/org/apache/hadoop/hbase/procedure2/NoNodeDispatchException.java
M hbase-procedure/src/main/java/org/apache/hadoop/hbase/procedure2/NoServerDispatchException.java
M hbase-procedure/src/main/java/org/apache/hadoop/hbase/procedure2/NullTargetServerDispatchException.java
Specializiations on FRDE so we can be more particular when we say there
was a problem.
M hbase-procedure/src/main/java/org/apache/hadoop/hbase/procedure2/RemoteProcedureDispatcher.java
Change addOperationToNode so we throw exceptions that give more detail
on issue rather than a mysterious true/false
M hbase-protocol-shaded/src/main/protobuf/MasterProcedure.proto
Undo SERVER_CRASH_HANDLE_RIT2. Bad idea (from HBASE-20173)
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/ServerManager.java
Have expireServer return true if it actually queued an expiration. Used
later in this patch.
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/AssignmentManager.java
Hide methods that shouldn't be public. Add a particular check used out
in unassign procedure failure processing.
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/MoveRegionProcedure.java
Check that server we're to move from is actually online (might
catch a few silly move requests early).
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/RegionStates.java
Add doc on ServerState. Wasn't being used really. Now we actually stamp
a Server OFFLINE after its WAL has been split. Means its safe to assign
since all WALs have been processed. Add methods to update SPLITTING
and to set it to OFFLINE after splitting done.
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/RegionTransitionProcedure.java
Change logging to be new-style and less repetitive of info.
Cater to new way in which .addOperationToNode returns info (exceptions
rather than true/false).
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/UnassignProcedure.java
Add looking for the case where we failed assign AND we should not
suspend because we will never be woken up because SCP is beyond
doing this for all stuck RPCs.
Some cleanup of the failure processing grouping where we can proceed.
TODOs have been handled in this refactor including the TODO that
wonders if it possible that there are concurrent fails coming in
(Yes).
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/procedure/ServerCrashProcedure.java
Doc and removing the old HBASE-20173 'fix'.
Also updating ServerStateNode post WAL splitting so it gets marked
OFFLINE.
A hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestServerCrashProcedureStuck.java
Nice test by Duo Zhang.
Signed-off-by: Umesh Agashe <uagashe@cloudera.com>
Signed-off-by: Duo Zhang <palomino219@gmail.com>
Signed-off-by: Mike Drob <mdrob@apache.org>
Adds new stripped-down, faster ByteBufferKeyValue comparator
(BBKV is the base Cell-type in hbase2). Creates an instance
of new Comparator each time we create new memstore rather
than use the universal CellComparator.
Remove unused and unneeded Interfaces from Cell base type.
- CacheStats won't generate NaN metrics.
- JSONBean class will serialize special floating point values as
"NaN", "Infinity" or "-Infinity"
Signed-off-by: Andrew Purtell <apurtell@apache.org>
Add method the CellComparator Interface. Add implementation to
meta comparator so we don't fall back to the default comparator.
Includes a nothing change to hbase-server/pom.xml just to provoke
build.
When we pread, we don't force the read to read all of the next block header.
However, when we get into a race condition where two opener threads try to
cache the same block and one thread read all of the next block header and
the other one didn't, it will fail the open process. This is especially important
in a splitting case where it will potentially fail the split process.
Instead, in the caches, we should only fail if the required blocks are different.
Signed-off-by: Andrew Purtell <apurtell@apache.org>
Change the MemStore size accounting so we don't synchronize across three
volatiles applying deltas. Instead:
+ Make MemStoreSize, a datastructure of our memstore size longs, immutable.
+ Undo MemStoreSizing being an instance of MemStoreSize; instead it has-a.
+ Make two MemStoreSizing implementations; one thread-safe, the other not.
+ Let all memory sizing longs run independent, untied by
synchronize (Huaxiang and Anoop suggestion) using atomiclongs.
+ Review all use of MemStoreSizing. Many are single-threaded and do
not need to be synchronized; use the non-thread safe counter.
TODO: Use this technique accounting at the global level too.
Add backoff when stuck in RegionTransitionProcedure, the subclass of
AssignProcedure and UnassignProcedure. Can happen when we go to
transition but the current Region state is not what we expect.
M hbase-procedure/src/main/java/org/apache/hadoop/hbase/procedure2/Procedure.java
Add doc on being able to suspend and wait on a timeout.
M hbase-protocol-shaded/src/main/protobuf/MasterProcedure.proto
Add 'attempt' counter so we can do backoff when we get stuck.
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/AssignProcedure.java
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/UnassignProcedure.java
Add persistence of new 'attempt' counter
M hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/RegionTransitionProcedure.java
Doc data members that are persisted by subclasses given this is 'odd'.
Add a counter for 'attempts' used when 'stuck' to implement backoff.
Add suspend with timeout when 'stuck'. Add callback when timeout is
exhausted which does wakeup of this procedure.
A hbase-server/src/test/java/org/apache/hadoop/hbase/master/assignment/TestUnexpectedStateException.java
Test of backoff.
Fix test that depended upon flush being slow and one family only.
Fix MemStoreSize compare to allow passing alternate implementation
(needed when IMC was no longer default everywhere).
HBase server versioned is checked after connecting to the server and then following operations are not allowed:
-fix, -fixAssignments, -fixMeta, -fixHdfsHoles, -fixHdfsOrphans, -fixTableOrphans, -fixHdfsOverlaps, -maxMerge
-sidelineBigOverlaps, -maxOverlapsToSideline, -fixSplitParents, -removeParents, -fixEmptyMetaCells
-repair, -repairHoles
* Make CleanerChore less chatty: move WARN message to DEBUG when we expect non-empty dirs
* Make CleanerChore less chatty: move IOE we'll retry to INFO
* CleanerChore should treat IOE for FileStatus as a failure
* Add tests asserting assumptions in above
Signed-off-by: Reid Chan <reidddchan@outlook.com>
Signed-off-by: Mike Drob <mdrob@apache.org>
Vasudevan
Michael Stack
Francis Liu
Ted Yu
Umesh Agashe
Mike Drob
Duo Zhang
David Manning
krish.dey
Guangxu Cheng
Allan Yang
Balazs Meszaros
Andrey Elenskiy
Andrew Purtell
Josh Elser
Sakthi
Nihal Jain
jingyuntian
brandboat
Wei-Chiu Chuang
meiyi
TAK LON WU
Xu Cang
Alex Leblang
huzheng
Reid Chan
Yu Li
Toshihiro Suzuki
zhaoyuan
Abhishek Singh Chouhan
Ted Yu
huzheng
Ankit Singhal
Sahil Aggarwal
Todd Lipcon
Sean Busbey
taiynlee
Guanghao Zhang
eric-maynard
Peter Somogyi
maoling
eshcar
Thiruvel Thirumoolan
Apekshit Sharma
Zach York
Guanghao Zhang
Chia-Ping Tsai
Rich Fecher
Huaxiang Sun
lujie
Ashish Singhi
BELUGA BEHR