Tags: svoflee/impala
Tags
Fix error handling in HdfsTableSink::Close() Update files open metric even if close fails. Protect against calling close multiple times. Change-Id: I0514665eca83bb7c67c8dfbce028c7afe4358b72 Reviewed-on: http://gerrit.ent.cloudera.com:8080/314 Reviewed-by: Alex Behm <alex.behm@cloudera.com> Tested-by: jenkins Reviewed-by: Lenni Kuff <lskuff@cloudera.com> (cherry picked from commit 0946dd7) Reviewed-on: http://gerrit.ent.cloudera.com:8080/316 Reviewed-by: Nong Li <nong@cloudera.com>
Implicit commands run by the shell should be delimited properly
Add lock in thread resource mgr to prevent using the ReaderContext af… …ter Unregister.
Make the shell work on SLES by adding pkg_resources.py from setuptools.
Impala code refresh v0.4 (pre-release version) -> v0.4 * Fix hdfs-scan-node deadlock. row_batches_lock_ should be taken after lock_. But in this bug, we took row_batches_lock_ first. * Fix race condition due to misuse of shared_ptr in StreamControlBlock. * Removing uses of perf-counters code (will not work on RHEL 5 kernel) * Fix race condition in opcode registry initialize. * Add CURRENT_TIMESTAMP() function as synonym for now() * Fix leak of open hdfs files during cancellation. * Fix race condition in scanners/scan-node. * Fix hang for queries when the limit is first reached by the coordinator. Remote fragments are waiting for the coordinator to pump rows but the coordinator decided it was done. This caused remote fragments to hang. The coordinator was in turn waiting for the remote fragments to finish. * Improved metrics for tracking impalad resource utilization. * Move sparrow/ to statestore/ and remove sparrow namespace * Added start/end query time to debug webpage. * Add build option to enable address sanitizer. * impala-shell calls PingImpalaService thrift API to verify the server is an impalad. * IMPALA-14: Files with .gz extension reported as 'not supported' * Use recompiled, relocatable static libraries when they are available