##############################################################################

    @RELEASE: 6.7-2

    ##############################################################################

    ==== CL 14581 ====
    @CHANGE: changed ("new") worker behavior when auto-mount drives are unmountable due to duplicate drives.

    Now, failed attempts to auto-mount a drive due to the drive letter already
    being in use will only generate a WARNING message in the workerlog, instead
    of rejecting the job and sending it back to the supe as "pending".

    ==== CL 14579 ====
    @CHANGE: add more useful info to print to the workerlog when a job is rejected due to duplicate drive mounting (attempt to mount to a drive letter that's already mounting something else)

    ==== CL 14574 ====
    @FIX: Secondary jobs were being dispatched even when supervisor_smart_share_mode is set to NONE

    ZD: 14613

    ==== CL 14528 ====
    @FIX: issue when modifying job's "env": "cwd", "umask", and "drivemap" are wiped-- additional fix to allow "env" modification of multiple jobs with a single call to qbmodify()

    See also CL14516.

    JIRA: QUBE-1161
    ZD: 14549

    ==== CL 14523 ====
    @CHANGE: upgraded supervisor's embedded Python to version 2.7.2 on Windows

    JIRA: QUBE-1164

    ==== CL 14518 ====
    @CHANGE: worker_boot_delay defaults to 10 seconds on workers running in service mode, on ALL platforms.

    JIRA: QUBE-989

    ==== CL 14516 ====
    @FIX: issue when modifying job's "env": "cwd", "umask", and "drivemap" are wiped

    JIRA: QUBE-1161
    ZD: 14549

    ==== CL 14514 ====
    @FIX: add agenda item (aka "work") status to print properly to the job's history log when it's recalled, because of the instance that's processing being migrated, interrupted, failed, killed, or blocked.

    There will be a line like the following in the .hst history log:

    [Sep 15, 2015 17:09:05] 495670145 work 45765 1 __QUBE_SYSTEM__@supervisor recalled in supervisor by user[] from host[supervisor] on host[shinyambp] (127.0.0.1)

    Note that this will also show, as expected, when a job instance reaches timeout (if specified) and "failed" by the system.

    JIRA: QUBE-829
    ZD: 13521

    ==== CL 14507 ====
    @FIX: issue where subst mounted local drives will disappear from Explorer after a job finishes on DU mode workers.

    @FIX: also fixed a bug where an already-mounted network/subst drives weren't being detected properly

    ZD: 14009
    JIRA: QUBE-1030

    ==== CL 14500 ====
    @FIX: issue where cmd* jobtype jobs fail when paths given to QB_CONVERT_PATH() include parentheses

    Note: problem was with the command-line tokenizer, QbExpressions::commandtokenize() routine, commonly used by all cmd* jobtypes, not respecting double-quoted and single-quoted strings.

    JIRA: QUBE-1139

    ==== CL 14479 ====
    @FIX: QB_CONVERT_PATH() runtime path conversion fails when the path to be converted contains parentheses

    ==== CL 14473 ====
    @FIX: Allow custom algorithms to decide how to preempt SmartShare secondary instances, or just default to using value set in supervisor_smart_share_preempt_policy.

    Custom algorithms may define a qb_preemptcmp_secondary() routine to control how secondary jobs are preempted.

    ZD: 14472
    JIRA: QUBE-1145

    ==== CL 14397 ====
    @FIX: performance tweak, cut down on the number of times backends and automated scripts fetch the supervisor config

    ==== CL 14360 ====
    @CHANGE: agenda-based job instance is immediately interrupted, even if the global preemption policy is set to passive, if it hasn't started processing an agenda item

    JIRA: QUBE-1077
    ZD: 14109

    ==== CL 14352 ====
    @FIX: added QB_FRAME_NUMBER, QB_FRAME_START, QB_FRAME_END, QB_FRAME_STEP, and QB_FRAME_RANGE to be defined in the environment just before a frame is processed

    ZD: 14203

    ==== CL 14326 ====
    @FIX: make appropriate invocation of approvemodify (qb_approvemodify() perl routine) for Custom Policy

    ZD: 14173
    JIRA: QUBE-1082

    ==== CL 14320 ====
    @FIX: catch case in checkUserPermission where traceback error "e" is not defined and an attempt is made to report the error message - occurs when user running the script is not a qube admin

    ==== CL 14305 ====
    @TWEAK: print queuing policy (Internal or custom/Perl) message to supelog

    ==== CL 14273 ====
    @FIX: properly report back failing status when an regex_error is matched early on, but then not found in the last pass through the logs.

    ==== CL 14207 ====
    @FIX: log sections that match an error regex from before an auto-retry are being scanned and matching for errors; now either "'qube! - retry/requeue" or "auto-retry" messages trigger a reset

    ==== CL 14204 ====
    @NEW: a script and modules to sync external 3rd-party license server counts with Qube's global resources
    @NEW: first external license server modules are for FLEXlm and sesinetd servers

    • No labels