Skip to content

Latest commit

 

History

History
365 lines (255 loc) · 9.93 KB

CHANGELOG.rst

File metadata and controls

365 lines (255 loc) · 9.93 KB
Raw

0.9.20161222

  • Bugfix: hotfix for tar_out issue #283, #284

0.9.20161221

  • Feature enhancement: Need for speed: compression of ansi codes #279
  • Feature enhancement: report version #276
  • Feature enhancement: some more fixes on help page, visible with ? #273
  • Bugfix: fix for Anonymisation feature temporarily unavailable until completely implemented (use --experimental) #170
  • Bugfix: fix for stacktrace while collecting anonymized sample (which by then worked fine otherwise) #280, #281
  • Bugfix: fix for when in filtered view, summary now reflects filtered jobs/cores #276
  • Bugfix: fix for svc.* no longer with Cyan color #278
  • Bugfix: fix-anonymize-SGE-and-optimisations #276
  • Bugfix: fix broken coloring #276

0.9.20161216

  • Feature enhancement: key H: highlight users and queues by string or regex #272
  • Feature enhancement: add -rr option: filter out unused core lines #270, #271
  • Feature enhancement: harmonize output filenames for window/full view and json exports #269
  • Bugfix: -r Option to remove empty core lines now works as advertised #267
  • Bugfix: fix for more arcane pbs node names in pbsnodes #266
  • Bugfix: fix for PBS plugin fails when jobs use non-consecutive CPUs #265
  • Bugfix: fix for adding -G flag under demo resulted in a crash, when getent was missing #262

0.9.20161207

  • Feature enhancement: online help available with ? ; #257
  • Feature enhancement: unhardwire __version__, make it visible across code; #259
  • Feature enhancement: Remove WORKDIR from banner, it was no longer serving a purpose; #243
  • Bugfix around missing core_job_map, when not remapping; #253
  • Bugfix unhardwire /tmp, make the choice of intermediate directory liberal; #254, #255
  • Bugfix under linux watch mode, fix watch issues of various types; #206, #248, #256
  • Bugfix around jobid consistency of input files in OAR - handle gracefully and report in debug log; #258

0.9.20161130

  • 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

0.8.9

  • 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

0.8.8

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

0.8.7

  • 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

0.8.6

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

0.8.5

  • initial support for data anonymisation
  • minor bugfixes

0.8.4

  • 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

0.8.3

  • 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

0.8.2

  • 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

0.8.1

  • Overwrote PyYAML dependency with custom YAML parser

0.8.0

  • Wrapping together support for PBS`, ``OAR, SGE

0.7.3

  • Support for OAR

0.7.2

  • Introduce support for SGE

0.7.1

  • Finalise support for PBS

0.7

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

0.6.7

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

0.6.6

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

0.6.5

Enhancements: - PBS now supported

0.6.4

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

0.6.3

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

0.6.2

Bugfixes: - WN matrix width bug ironed out.

0.6.1

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

0.5.2

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

0.5.1

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

0.5.0

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

0.4.1

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

0.4.0

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

0.3.0

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

0.2.9

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)

0.2.8

Bugfixes: - colour implementation for all of the tables

0.2.7

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

0.2.6

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

0.2.5

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

0.2.4

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

0.2.3

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)

0.2.2

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

0.2.1

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

0.2.0

Bugfixes: - unix accounts are now correctly ordered

0.1.9

Bugfixes: - All CPU lines displaying correctly

0.1.8

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

0.1.7

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

0.1.6

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

0.1.5

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

0.1.4

Bugfixes: - some “wiremelting” concerning the save directory

0.1.3

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

Enhancements:
- Now each state is saved in a separate file in a results folder

0.1.2

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

0.1.1

Enhancements: - changed implementation in get_state()

0.1.0

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