MapR Patch Release - July, August, September 2016

Document created by Rachel Silver Employee on Oct 31, 2016Last modified by lnelsonking on Nov 15, 2016
Version 3Show Document
  • View in full screen mode

Get Notified of New Release & Patch Announcements

 

MapR Patch Releases for July-September 2016

The following list of patches are interim patches released between maintenance release cycles. Each patch released (mid-month) is cumulative in nature and includes prior patches released for the branch used.

 

 

5.2.0

 

Read the MapR 5.2.0 Release Notes for details

 

Bug 11349 - Hoststats did not work on POSIX edge node.

 

Bug 23652 - The POSIX loopbacknfs client does not automatically refresh renewed service tickets.

 

Bug 23975 - MFS fails to start for certain Docker containers (In version 5.1).

 

Bug 23981 – Excessive CPU utilization encountered during remote writes as MFS was conducting CRC checksum in the MFS thread.

 

Bug 24022 - Hung thread encountered during the mirroring of a volume on a container which did not have a master container.

 

Bug 24053 - During client initialization, the client crashed if there was an error during initialization.

 

Bug 24119 - Too much memory allocated to MFS as Warden adjusts the FileServer (MFS) and Node Manager (NM) memory incorrectly when NM and TaskTracker (TT) are on the same node.

 

Bug 24159 - Update timestamp (mtime or ctime) does not display nanoseconds when a hard link is created from the FUSE mount point or updated.

 

Bug 24280 - Running the maprcli dashboard info command occasionally throws a TimeoutException error.

 

Bug 24232 - Users are prevented from accessing files in the parent directory when those files were created with stale chunk IDs.

 

Bug 24315 – On an NFS client, an incorrect amount of data may be read when using the dd command with iflag=direct. 

 

Bug 24324 - The “disk not found” error was encountered when using a script targeting disks across all instances of the fileserver process.

 

Bug 24392 - When upgrading from Hadoop 2.5.1 to Hadoop 2.7.1, the ResourceManager fails to recover the state, as version 2.7.1 uses a new state store data format.

 

Bug 24413 - CLDB crashes when volume replication is greater than 3.

 

Bug 24446 -The getattr call sometimes returns a stale mtime value due to incorrect attribute cache handling in NFS server.

 

Bug 24562 - CLDB (container location database) performance suffers due to the Warden allocation of the CLDB service as lower CPU priority.

 

Bug 20498 and 24143 - JobTracker service encounters lock contention and becomes unresponsive as a result of multiple job configuration objects loaded via TaskTracker.

 

Bug 23530 and 23260 - In certain cases, containers were stuck in an incomplete state when either the request for role or response from the replicate state command was lost.

 

Bug 23876 – When the same node is added to the replica chain twice, the resync request is blocked.

 

Bug 23931 - Nested user queues do not inherit labels from their parent queue.

 

Bug 24054 - When MapR-DB is used for time series data, a large number of tablets are generated with empty partitions. This can be merged into single tablet. However, in this case, the single tablet will have have thousands of empty partitions. Having a single tablet with thousands of empty partitions may result in a degradation of MapR-DB performance.

 

Bug 24056 - Client crashes while printing error message due to lack of arguments passed to the gtrace function.

 

Bug 24059 - Inability to retrieve the list of containers unaware of the rack.

 

Bug 24063 - During mirroring, volume property update messages logged (in cldb.log) every 5 seconds due to log level setting of INFO.

 

Bug 24097 - CLDB threads busy/unresponsive as the retrieval of the ZooKeeper address for every Hadoop command required a scan of the complete cluster information.

 

Bug 24140 - The MapR Client throws an assert and the application crashes if the MAPR_TICKETFILE_LOCATION points to an incorrect location when attempting to access a secure cluster from Windows.

 

Bug 24170 -  When using applications like YARN, that require a ZooKeeper string, on a MapR FileClient version 5.1.0 or higher, connections to clusters running older versions may fail.

 

Bug 24190 – During replication, the destination table takes a long time to catch up with the primary table when the table count is large (greater than 10,000).

 

Bug 24238 - MFS crashed with asserts in Unlink due to an issue with rename when the target file has multiple hard links.

 

 

 

5.1.0

 

Read the MapR 5.1.0 Release Notes for details

 

Bug 23652 - The POSIX loopback NFS client does not automatically refresh renewed service tickets.

 

Bug 24053 - During client initialization, the client crashed if there was an error during initialization.

 

Bug 24057 - Only SPs in average and above buckets are considered for balancing when disk balancer had to spread the containers for optimal SP utilization.

 

Bug 24119 - Too much memory allocated to MFS as Warden adjusts the FileServer (MFS) and Node Manager (NM) memory incorrectly when NM and TaskTracker (TT) are on the same node.

 

Bug 24159 - Update timestamp (mtime or ctime) does not display nanoseconds when a hard link is created from the FUSE mount point or updated.

 

Bug 24232 - Users are prevented from accessing files in the parent directory when those files were created with stale chunk IDs.

 

Bug 24324 - The “disk not found” error encountered when using a script targeting disks across all instances of the fileserver process.

 

Bug 24280  - Running the maprcli dashboard info command occasionally throws a TimeoutException error.

 

Bug 20498 and 24143 - JobTracker service encounters lock contention and become unresponsive as a result of multiple job configuration objects loaded via TaskTracker.

 

Bug 23776 Java API client crashed as the closed_flag in Inode close() method does not set to true during an error (like EACCES).

 

Bug 23931 - Nested user queues do not inherit labels from their parent queue.

 

Bug 23933 - When an OJAI document is created by parsing a JSON string containing a List of Maps, an extraneous null element is added to the list in the resultant document.

 

Bug 24018 - YARN jobs failed with "Rename cannot overwrite non empty destination directory.”

 

Bug 24022 - Hung thread encountered during the mirroring of a volume on a container which did not have a master container.

 

Bug 24025 - When the HistoryServer read the job history file for a job that was not initialized correctly, it read "-" as a delimiter. This caused the job start time to have an empty value. As a result, the following warning displayed: <DATE> <TIME> WARN org.apache.hadoop.mapreduce.v2.jobhistory.FileNameIndexUtils: Unable to parse start time from job history file.

 

Bug 24050 - The output of the MapReduce LineRecordReader function occasionally had the following issues: With the multibyte record delimiter, records were dropped due to incorrect split processing. With the multibyte record delimiter, duplicate records were produced. With the custom delimiter, incomplete records were read. With the custom and default delimiter, incorrect key/position information was used for uncompressed input.

 

Bug 24054 - When MapR-DB is used for time series data, a large number of tablets are generated with empty partitions. This can be merged into a single tablet. However, in this case, the single tablet will have thousands of empty partitions. Having a single tablet with thousands of empty partitions may result in the degradation of MapR-DB performance.

 

Bug 24063 - During mirroring, volume property update messages logged (in cldb.log) every 5 seconds due to log level setting of INFO.

 

Bug 24097 - CLDB threads busy/unresponsive as the retrieval of the ZooKeeper address for every Hadoop command required a scan of the complete cluster information.

 

Bug 24140 - The MapR Client throws an assert and the application crashes if the MAPR_TICKETFILE_LOCATION points to an incorrect location when attempting to access a secure cluster from Windows.

 

Bug 24170 -  When using applications like YARN, requiring a ZooKeeper string, on a MapR FileClient version 5.1.0 or higher, connections to clusters running older versions may fail.

 

Bug 24190 – During replication, the destination table takes a long time to catch up with the primary table when the table count is large (greater than 10,000).

 

Bug 22368 – Indefinite looping of mrconfig info dumpcontainers command.

 

Bug 23331 - When re-reading large files, cache misses were seen despite warm cache.

 

Bug 23353 - Spark 1.5.2 job server core dumps when running with MapR 5.1 client.

 

Bug 23488 - Volumes with a replication factor below 3 and no minimum number of replication defined, the replication manager sets the minimum replication value to 1 and containers with only one copy are not re-replicated.

  

Bug 23544 - Incorrect values are returned for Name Container Replication Factor. Namespace and Dataspace replication can be independently specified. However, if Namespace replica factors are not set explicitly, they assume the same values as that of data spacefor both desired and min.

 

Bug 23569 - The files under the /tmp/mapr-hadoop/mapred/local/toBeDeleted/ directory are not deleted when the TaskTracker service is restarted or when it reconnected with Jobtracker.

 

Bug 23596 - When Fastfailover is enabled and star replica is implemented, the system may encounter Incomplete replication under certain scenarios. 

 

Bug 23629 - While allocating a large number of inodes during the resynchronization of containers, the source container times out if the destination container does not respond within 5 minutes.

 

Bug 23676 - When File Access Control Expressions (ACEs) are set on a symbolic link, flags are set to indicate ACEs are set on the symlink file itself.

 

Bug 23687 – During the use of commands hdfsOpenFile(), hdfsWrite(), hdfsFlush(), and/or hdfsCloseFile() on a file in which the executing user does not have the correct permissions for the operation, 0 was returned instead of the right error code.

 

Bug 23715 - The MFS C and Java APIs do not return the requested number of bytes.

 

Bug 23725 - CPU spin was observed during CLDB failover on clusters with more than 50,000 volumes.

 

Bug 23745 - Pig jobs failing, on a secure cluster, when zero-configuration Resource Manager HA failed to handle a non MapR-FS job configuration object.

 

Bug 23762 - Error hdfs.h did not expose hdfsExists2() returned When compiling code with hdfsExists2().

 

Bug 23791 - The ResourceManager UI does not display details (shows Processing) for task attempts until jobs complete.

 

Bug 23795 - Storage pools going offline frequently with CRC errors due to: 

  • NFS server encrypting files when network-encryption bit is not set.
  • During replication, source retransmitting without verifying CRC.

Bug 23944 - When a local write times out with the ETIMEDOUT error, in certain cases, the NFS server re-uses shared pages, before mfs releases those pages, resulting in mfs crash.

 

Bug 24008 – On an NFS node, a CLDB crash occurs when attempting to get the port from the IP list of a server that did not send the IPs.

 

 

5.0.0

 

Read the MapR 5.0.0 Release Notes for details

 

Bug 23652 - The POSIX loopback NFS client does not automatically refresh renewed service tickets.

 

Bug 24053 - During client initialization, the client crashed if there was an error during initialization.

 

Bug 24119 - Too much memory allocated to MFS as Warden adjusts the FileServer (MFS) and Node Manager (NM) memory incorrectly when NM and TaskTracker (TT) are on the same node.

 

Bug 23944 - When a local write times out with the ETIMEDOUT error, in certain cases, the NFS server re-uses shared pages, before MFS releases those pages, resulting in MFS crash.

 

Bug 24022 - Hung thread encountered during the mirroring of a volume on a container which did not have a master container.

 

Bug 24063 - During mirroring, volume property update messages logged (in cldb.log) every 5 seconds due to log level setting of INFO.

 

Bug 24140 - The MapRClient throws an assert and the application crashes if the MAPR_TICKETFILE_LOCATION points to an incorrect location when attempting to access a secure cluster from Windows.

 

Bug 22368 – Indefinite looping of mrconfig info dumpcontainers command.

 

Bug 23488 - With Volumes with a replication factor below 3 and no minimum number of replication defined, the replication manager sets the minimum replication value to 1 and containers with only one copy are not re-replicated.

 

Bug 23544 - Incorrect values are returned for Name Container Replication Factor. Namespace and Dataspace replication can be independently specified. However, if Namespace replica factors are not set explicitly, they assume the same values as that of data spacefor both desired and min.

 

Bug 23629 - While allocating a large number of inodes during the resynchronization of containers, the source container times out if destination container does not respond within 5 minutes.

 

Bug 23715 - The MFS C and Java APIs did not return the requested number of bytes.

 

Bug 23795 - Storage pools going offline frequently with CRC errors due to: 

  • NFS server encrypting files when network-encryption bit is not set.
  • During replication, source retransmitting without verifying CRC.

 

Bug 23799 – When encountering an error, container resync work area freed while the inode resync work area is still referring to the container resync work area.

 

Bug 23876 – When the same node is added to the replica chain twice, the resync request is blocked.



 

4.1.0

 

Read the MapR 4.1.0 Release Notes for details

 

Bug 24053 - During client initialization, the client crashed if there was an error during initialization.

 

Bug 21930 - Hadoop2.5.1 -RM GUI showed values that are too high for reserved memory and cores; actual values were substantially lower.

 

Bug 24022 - Hung thread encountered during the mirroring of a volume on a container which did not have a master container.

 

Bug 24063 - During mirroring, volume property update messages logged (in cldb.log) every 5 seconds due to log level setting of INFO.

 

Bug 24140 - The MapRClient throws an assert and the application crashes if the MAPR_TICKETFILE_LOCATION points to an incorrect location when attempting to access a secure cluster from Windows.

 

Bug 22124 - CLDB crashes due to loopback NFS server and POSIX client simultaneously participating in VIP rebalancing.

 

Bug 22368 – Indefinite looping of mrconfig info dumpcontainers command.

 

Bug 23629 - While allocating a large number of inodes during the resynchronization of containers, the source container times out if destination container does not respond within 5 minutes.

 

Bug 23715 - The MFS C and Java APIs did not return the requested number of bytes.

 

Bug 23745 - Pig jobs failing, on a secure cluster, when zero-configuration Resource Manager HA failed to handle a non MapR-FS  job configuration object.

 

Bug 23799 - When there is an error, the container resync work area was freed, but the inode resync work area was still referring to the container resync work area.

 

Bug 23876 – When the same node is added to the replica chain twice, the resync request is blocked.

 

Bug 23946 - Memory limit update does not take effect when the Xmx parameter for mapreduce.reduce.java.opts is increased to a value greater than 2GB.

 

 

 

4.0.2

 

Read the MapR 4.0.2 Release Notes for details

Bug 21930 - Hadoop2.5.1 -RM GUI showed values that are too high for reserved memory and cores; actual values were substantially lower.

 

Bug 23852 - The ResourceManager UI and Warden do not use the same logic to determine the list of running NodeManager nodes.

 

Bug 24022 - Hung thread encountered during the mirroring of a volume on a container which did not have a master container.

 

Bug 24025 - When the HistoryServer read the job history file for a job that was not initialized correctly, it read "" as a delimiter. This caused the job start time to have an empty value. As a result, the following warning displayed: <DATE> <TIME> WARN org.apache.hadoop.mapreduce.v2.jobhistory.FileNameIndexUtils: Unable to parse start time from job history file.

 

Bug 24050 - The output of the MapReduce LineRecordReader function occasionally had the following issues:

  • With the multibyte record delimiter, records were dropped due to incorrect split processing.
  • With the multibyte record delimiter, duplicate records were produced.
  • With the custom delimiter, incomplete records were read.
  • With the custom and default delimiter, incorrect key/position information was used for uncompressed input.

 

Bug 24059 - Inability to retrieve the list of containers unaware of the rack.

 

Bug 24140 - The MapRClient throws an assert and the application crashes if the MAPR_TICKETFILE_LOCATION points to an incorrect location when attempting to access a secure cluster from Windows

 

Bug 24063 - During mirroring, volume property update messages logged (in cldb.log) every 5 seconds due to log level setting of INFO.

 

Bug 17642 - Resync failed with ENOSPACE error. Process attempted to reserve more space than required for resync to complete.

 

Bug 22124 - CLDB crashes due to loopback NFS server and POSIX client simultaneously participating in VIP rebalancing.

 

Bug 22368 – Indefinite looping of mrconfig info dumpcontainers command.

 

Bug 23541 - A ddlopen of libmapr_pam.so using immediate symbol resolution throws an undefined symbol error.

 

Bug 23545 - MapReduce Application Master does not honor the binding port range specified through the yarn.app.mapreduce.am.job.client.portrange property.

 

Bug 23569 - The files under the /tmp/mapr-hadoop/mapred/local/toBeDeleted/ directory are not deleted when the TaskTracker service is restarted or when it reconnected with Jobtracker.

 

Bug 23591 - The deletion of snapshots delayed after commands to delete snapshots are issued, while deletions are recorded in the CLDB log before they are performed.

 

Bug 23629 - While allocating large number of inodes during the resynchronization of containers, the source container times out if destination container does not respond within 5 minutes.

 

Bug 23715 - The MFS C and Java APIs did not return the requested number of bytes.

 

Bug 23745 - Pig jobs failing, on a secure cluster,  when zero-configuration Resource Manager HA failed to handle a non MapR-FS  job configuration object.

 

Bug 23799 - When there is an error, the container resync work area was freed, but the inode resync work area was still referring to the container resync work area.

 

Bug 23876 – When the same node is added to the replica chain twice, the resync request is blocked.

 

Bug 23922 - The ResourceManager fails to start when FileSystemRMStateStore contains invalid application states.

 

Bug 23944 - When a local write times out with the ETIMEDOUT error, in certain cases, the NFS server re-uses shared pages, before mfs releases those pages, resulting in mfs crash.

 

Bug 24034 - RACK violation messages are continuously printed to cldb.log. This issue occurs since the Replication Manager prints queue stats every two minutes and Containers having Rack Violation are fixed after 12 hours.

Attachments

    Outcomes