XtreemFS Changelog

New in version 1.5.1

April 21st, 2015
  • Hadoop Adapter supports Hadoop-2.x and other applications running on the YARN platform (tested with Apache Spark and Apache Flink), new OSD selection policy that imitates the HDFS placement behavior.
  • Consistent adding and removing replicas for R/W replication.
  • xtfs_scrup now replaces failed replicas using the R/W replication policy.
  • Improvements of the SSL mode: the used SSL/TLS version is selectable, strict certificate chain checks are possible, the SSL code on client and server side was improved.
  • Support for PKCS12 certificate containers is remove from the Windows client (did not work in previous release).
  • mount.xtreemfs accepts all parameters to be passed in the form -o option=value. If XtreemFS is mounted via /etc/fstab, mount options are passed to the XtreemFS client.
  • Dockerfiles for the XtreemFS services are available at https://github.com/xtreemfs/xtreemfs-docker.
  • A Vagrantfile is available in the XtreemFS repository to ease development and testing for new contributors.
  • Fixed snapshots in the case of deleted files.
  • Snapshots can be deleted even after inactivating snapshots for a volume (issue 290).
  • Initial version of an LD_PRELOAD based client that bypasses FUSE (experimental).
  • Initial implementation of volume quotas (checked only while opening files by the MRC).
  • OSDs can report their health to the DIR (e.g. based on SMART values), which aggregates this information in the web interface. A new OSD selection policy sorts out unhealthy OSDs.
  • Improved automatic testing: introduced Travis-CI, migrated all JUnit tests to version 4, improved reliability of nightly integration tests.
  • Added new benchmark tool xtfs_benchmark that is based on the JAVA libxtreemfs.
  • Many small improvements and fixes.

New in version 1.5 (March 14th, 2014)

  • Improved Hadoop Adapter e.g., added support for multiple volumes and read and write buffer to speed up performance.
  • Support for SSDs: Support for multiple OSD storage threads to increase parallelism.
  • Status webpage as part of the DIR webinterface for replicated files to show current primary and backup replicas.
  • Multi-homing support. XtreemFS can now be made available for multiple networks and clients will pick the correct address automatically.
  • Support for multiple OSDs per machine (e.g. one for each disk) through the xtreemfs-osd-farm init.d script.
  • Fixed major issues in Read-Only and Read/Write file replication.
  • Reduced metadata cache default timeout from 120 to 10 seconds. Disable it completely with --metadata-cache-size=0.
  • Upgraded to Google Protobuffers 2.5 which improves performance.
  • Windows Client: Fixed several issues (e.g., 286, 299) and updated the CbFS driver to the latest version.
  • XtreemFS also compiles under buildroot now.
  • Fixed several issues around SSL (e.g., 263, 281).
  • Fixed many small issues e.g., reported volume capacity was inaccurate (issue 296).
  • Other fixed issues: 267, 268, 272, 274, 276, 277, 278, 283, 287, 288, 292, 293, 294.
  • Improved general stability (e.g., see issues 297, 301).
  • xtfs_scrub now repairs replicas with an invalid checksum automatically.
  • Disabled deferred deletion of file data on the OSDs.
  • Improved test coverage of nightly tests and added missing unit tests and a Valgrind leak check for C++ unit tests.

New in version 1.3.1 (November 12th, 2012)

  • Added 'xtfs_remove_osd' utility to remove OSDs from an XtreemFS installation. All files on the OSD to be removed will be relocated to different OSDs.
  • Added first support for asynchronous writes in the client. Mounting with '--max-writeahead' and '--max-writeahead-requests' allows the user to specify the maximum number of bytes and requests to be sent to an OSD before receiving an acknowledgment. However, this feature currently does not work together with replication and retries.
  • Added monitoring support to servers. Various characteristics are accessible now via SNMP, such as data transfer statistics, memory usage, numbers of files and directories, etc. A Ganglia plugin makes it possible to visualize them.
  • improved stability of MRC/DIR replication
  • improved the behavior of the read-write replication under heavy load by internally limiting the length of request queues
  • made several usability changes to xtfsutil
  • added '/babudb' subpage to MRC/DIR status page that shows internal database statistics
  • added Gentoo overlay
  • installing XtreemFS now creates symlinks '/sbin/mount.xtreemfs' and '/sbin/umount.xtreemfs' in order to support XtreemFS mounting via fstab (however, not all client options are available in the fstab yet, see issue 205)
  • the replication policy "WaRa" was renamed to "WaR1"
  • fixed issues 196, 198-204, 206-209, 211-218, 220, 222
  • fixed a wide range of issues w/ read-only and read-write replication of files
  • fixed issues w/ several MRC calls when admin passwords were specified (a consequence of this is that an XtreemFS 1.3.0 client can no longer mount admin_password protected MRCs; upgrade the client to 1.3.1 to fix this)
  • fixed various client issues, such as deadlocks and memory leaks