• Summary in the 1st section now shows Total:, Up:, Free: nodes
• Added a nodes column per user in the 3rd section where user info is presented
• Added -R feature to replay last frames; recording is automatic!
• Added setup.py, pip install --user qtop should work
• Added --version, in case anyone wondered about it
• Bail out if python version is <2.5
• Supplied qtop launcher script, for better system integration
• Create timestamped intermediate files, to improve ability to look back
• Bugfix to avoid /tmp getting crowded with temporary files
• Bugfix for grepping and pipelining qtop's output (1st attempt to treat #206)
• Bugfixes on PBS (jobid arrays, node states etc)
• Bugfixes on SGE (job states added, all jobs now visible, userid is correct)
• Bugfixes on OAR (nodes per user are also visible)
• Converted documentation files to .rst format

• watch mode (-w)
  • ((s)) node sorting with custom presets/user-inserted RegEx sorts
  • ((f)) node filtering
  • ((t)) matrix transposition
  • ((F)) toggle full nodename display/numbering
  • ((m)) toggle coloring code (user-id based/queue name-based)
• queue info can now be colored, three different queues with the same initials can be colored distinctly
• a good first amount of documentation
• small bugfixes

• created a shiny new demo out of a demo mini-grid scheduler (!)

• added support for queuename display for OAR and PBS (SGE already there!)
• overspill feature (aka oversubscribe/overcommitment) for SGE systems
• strict checking cmdline var (will compare reported nr. of jobs against displayed/found jobs)
• id column is now more readable (put in brackets)
• fixed long-standing bug that would report incorrectly the queues of jobs in a node, in some cases
• restored deprecated document file deletion
• huge refactoring of the codebase

• Support for python 2.6 for early RHEL6/Centos6/ScientificLinux6 distros
• Added tarball creation for better user reporting
• Added Viewport class
• Several bugfixes

• initial support for data anonymisation
• minor bugfixes

• Watch replacement with simple pager included (with full color functionality, compatible with older RHEL6 systems)
• GECOS field completed by a less “intruding” command
• Numerous enhancements and bug fixes

• The worker node occupancy table can now be viewed horizontally (transposed)
• Custom conf files createable by users
• Filter/select nodes by name/regex
• Numerous enhancements and bug fixes

• Ability to select which of the three qtop parts to display
• GECOS field implemented
• States can be assigned more than one lines
• New node line displays the queue the job belongs to
• (trivial) auto-detection of batch-system
• Numerous enhancements and bug fixes

• Overwrote PyYAML dependency with custom YAML parser

• Wrapping together support for PBS, OAR, SGE

• Support for OAR

• Introduce support for SGE

• Finalise support for PBS

Enhancements: - Input files are now using YAML for dumping and loading

Enhancements: - created yaml files now have the pid appended to the filename - pbs-related functions (which create the respective yaml files) have moved to a dedicated module - took out state_dict[‘highest_core_busy’], seemed useless (and unused)

Bugfixes: - a separate read_qstatq_yaml function added, for consistency (removed from qstatq2yaml) - change qstatq_list from list of tuples to list of dictionaries - offline_down_nodes was moved from pbs.pbsnodes2yaml to read_pbsnodes_yaml

Bugfixes: - got rid of all global variables (experimental)

Enhancements: - PBS now supported

Bugfixes: - lines that don’t contain any actual core are now not printed in the matrices.

Enhancements: - optional stopping of vertical separators (every ‘n’ position for x times) - additional vertical separator in the beginning

Bugfixes: - WN matrix width bug ironed out.

Enhancements: - Custom-cut matrices (horizontally, too!), -o switch

Enhancements: - Custom-cut matrices (vertically, not horizontally), width set by user.

Enhancements: - If more than 20% of the WNs are empty, perform a blind remap. - Code Cleanup

Bugfixes: - Major rewrite of matrices calculation fixed

New features: - true blind remapping !!

Enhancements: - exotic cases of very high numbering schemes now handled - more qstat entries successfully parsed - case of many unix accounts (>62) now handled

Bugfixes: - now understands additional probable names for pbsnodes,qstat and qstat-q data files

Bugfixes: - corrected colorless switch to have ON/OFF option (default ON) - qstat_q didn’t recognize some faulty cpu time entries - now descriptions are in white, as before.

Enhancements: - Queues in the job accounting summary section are now coloured

Enhancements: - command-line arguments (mostly empty for now)! - non-numbered WNs can now be displayed instead of numbered WN IDs

New features: - implement colorless switch (-c)

Bugfixes: - fixed issue with single named WN - better regex pattern and algorithm for catching complicated numbered WN domain names

New features: - handles cases of non-numbered WNs (e.g. fruit names) - parses more complex domain names (with more than one dash)

Bugfixes: - correction in WN ID numbers display (tens were problematic for larger numbers)

Bugfixes: - colour implementation for all of the tables

Bugfixes: - Exiting when there are two jobs on the same core reported on pbsnodes (remapping functionality to be added) - Number of WNs >1000 is now handled

Bugfixes: - fixed some names not being detected (%,= chars missing from regex)

Enhancements: - changed name to qtop, introduced configuration file qtop.conf and colormap file qtop.colormap

New features: - Working Cores added in Usage Totals - map now splits into two if terminal width is smaller than the Worker Node number

Enhancements: - implemented some stuff from PEP8 - un-hardwired the file paths - refactored code around cpu_core_dict functionality (responsible for drawing the map)

Bugfixes: - corrected regex search pattern in make_qstat to recognize usernames like spec101u1 (number followed by number followed by letter) now handles non-uniform setups - R + Q / all: all did not display everything (E status)

Enhancements: - masking/clipping functionality (when nodes start from e.g. wn101, empty columns 1-100 are ommited)

Enhancements: - Hashes displaying when the node has less cores than the max declared by a WN (its np variable)

Bugfixes: - unix accounts are now correctly ordered

Bugfixes: - All CPU lines displaying correctly

Enhancements: - unix account id assignment to CPU0, 1 implemented

Enhancements: - ReadQstatQ function (write in yaml format using Pyyaml) - output up to Node state!

Bugfixes: - ReadPbsNodes function (write in yaml format using Pyyaml)

Bugfixes: - implemented saving to 3 separate files, QSTAT_ORIG_FILE, QSTATQ_ORIG_FILE, PBSNODES_ORIG_FILE

Bugfixes: - some “wiremelting” concerning the save directory

Bugfixes: - fixed tabs-to-spaces. Formatting should be correct now.

Enhancements: - script reads qtop-input.out files from each job and displays status for each job

Enhancements: - changed implementation in get_state()

Enhancements: - just read a pbsnodes-a output file and gather the results in a single line