- 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>
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>
* 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>
Conflicts:
hbase-server/src/main/java/org/apache/hadoop/hbase/master/cleaner/CleanerChore.java
hbase-server/src/test/java/org/apache/hadoop/hbase/master/cleaner/TestCleanerChore.java
hbase-server/src/test/java/org/apache/hadoop/hbase/util/TestFSUtils.java
* in a post-step, build status can either be "null" or "SUCCESS" to indicate success.
* before we do an scm checkout for stages that post to the comment, set a default "we failed ¯\_(ツ)_/¯" comment.
Signed-off-by: Michael Stack <stack@apache.org>
Signed-off-by: Mike Drob <mdrob@apache.org>
* Ensure Jenkins steps that invoke bash inline set -e
* machine stats script should check that passed directory will work
Signed-off-by: Michael Stack <stack@apache.org>
* add a new test to build the refguide specifically instead of site
* check for changes to src/main/asciidoc or src/main/xslt and run that test and only that test
* check for changes to the hbase-default.xml file and build the refguide if found (but maybe other tests too)
* fallback to relying on the yetus default for other changes
* fix some missing start_clock entries that cause longer-than-actual reported test time.
Signed-off-by: Mike Drob <mdrob@apache.org>
* makes commands always pass any results back to hirb
* print warning if hirb is given the --return-values flag
* add some docs on how to avoid the console clutter that HBASE-15965 sought to address
* add an upgrade section note about this change.
* cleanup where the get_splits command does its printing so there's a building block that doesn't print
* some rubocop suggested tweaks and opt-out for classlength check on table and shell classes.
Signed-off-by: Mike Drob <mdrob@apache.org>
Conflicts:
src/main/asciidoc/_chapters/upgrading.adoc
* backport leaves off refguide change about upgrading.
Conflicts:
hbase-shell/src/main/ruby/hbase/table.rb
hbase-shell/src/main/ruby/shell/commands/get_splits.rb
When prefetch on open is specified, there is a deadlocking case
where if the prefetch is cancelled, the PrefetchExecutor interrupts
the threads if necessary, when that happens in FileIOEngine, it
causes an ClosedByInterruptException which is a subclass of
ClosedChannelException. If we retry all ClosedChannelExceptions,
this will lock as this access is expected to be interrupted.
This change removes calling refreshFileConnections for
ClosedByInterruptExceptions.
Signed-off-by: Andrew Purtell <apurtell@apache.org>