2019.01 Update 6 Release

New Features and Enhancements

The following new features and enhancements were introduced this software release:

Product Internal Number Case Number Description
All VOV-9777 24039 Added "Last Dispatch" into the page jobqueue?page=buckets&. Prior to fix, it was showing "age", which is confusing as it gets updated on submission and job dispatch.
Accelerator VOV-11617   Added support for SIGTSTP to vtool.

When vtool receives SIGTSTP, it releases its license tokens and suspends itself. Upon receiving SIGCONT, vtool resumes and eventually checks out the tokens again.

Accelerator VOV-10801 24711 Added support for keeping an interactive job's (nc run -I, -Ir, etc.) network connection with the remote job alive by periodically "pinging" the remote job. This option is enabled by setting the VOV_INTERACTIVE_PING environment variable to the desired ping interval (timespec format).
Accelerator VOV-11740 25241 Duplicate quantitative resource requests are no longer rejected when using the -r+ submission option. For consumable resource requests, the sum duplicate values is used. For all others, the maximum value is used.
"-r+ RAMFREE#100 -r+ RAMFREE#200" = RAMFREE#200
"-r+ RAM/100 -r+ RAM/200" = RAM/300

Note that nc info shows the resource string that was specified in the submission. The solution field and property of the job will contain the final result.

Accelerator VOV-11025 23337, 24800 Enhance Accelorator's tracking of the user processes that are part of a job by using UNIX session IDs and also by examining the process environment for the VOV_JOBID variable to associate it with the right NC job.
Accelerator VOV-11640 24715 Fixed issue where more than 10 rows of running jobs cannot be exported as CSV.
Accelerator VOV-11694   Fixed issue with parsing of -dpresN option in nc run
Accelerator VOV-5418 21072 In the output of the nc info command, if the slave for the job being inspected has a current status of SICK, the status will be indicated in parentheses after the name of the slave.
Accelerator VOV-11351 25139 The -dpres+ option is now supported in DP jobs.
Accelerator VOV-11155   The ALLPIDS property has been added to the online help.
Allocator VOV-9994 24210 Added an option to retain resource criticality up to a given number of cycles, which is configurable in vovlad/config.tcl using the variable MQ(criticalResCycle). By default this feature is disabled and will be enabled only for positive numbers.
Allocator VOV-10657 24264 Modified the behavior when only one queue (or a subset of queues) requires a shared license, such that leftover licenses are allocated to all queues based on weights. The earlier behavior was to allocate all available licenses to only the queue(s) that requires the shared license.
Allocator VOV-11540 25227 There is a change in how Allocator allocates tokens when a resource is critical on some queues, but not on others. Allocator will now try to allocate tokens to a queue where the resource is critical, in preference to a queue where it is not critical even if this leads to the allocation on the non-critical queue falling below the level needed to run all the queued jobs.
FlowTracer VOV-11793 25217 A new server parameter seatlic_max has been added that places an upper limit on the number of seat license features that can be used by a project.
The parameter can be set either in policy.tcl:
set config(seatlic) <max-seat-features>
or as a server configuration:
vtk_server_config seatlic_max <max-seat-features>

By default the value is unlimited.

FlowTracer VOV-12004 CS0120860 Set the provided shell variable value to zero, (export VOV_LICENSE_SHOW_VIOLATION=0), to suppress the license violation preamble when invoking vovsh commands.
FlowTracer VOV-12356   FlowTracer local resources may be created/updated using the vovresourcemgr command.
vovresourcemgr create foo -max 10 -local, vovresourcemgr set foo -max 10 -nolocal
Monitor VOV-11750 120618 Fixed an issue in Monitor where the Admin > System > Security page would display an error (can't read "policyInfo(readonlyPort)": no such element in array...) in some cases.

Resolved Issues

The following issues were resolved in this release.

Product Internal Number Case Number Description
All VOV-12236   Fixed an issue leading to the following server logs:
"vovserver(3260) ERROR May 24 00:00:53 
Found host with different ip: instead of [host:903]"
All VOV-11253 25045 Improved the error messaging for the vtk_resourcemap_forget API command.
Accelerator, Monitor VOV-11487   A bug fix in 2019.01u5 could result in issues when starting a server with the same requested port id specified in rapid succession, or when requesting a range of ports such as -port 500:501:502.
Accelerator VOV-11223   A typo in the example of the vovreconciled/config.tcl in the Altair Accelerator Administrator's Guide has been fixed.
Accelerator VOV-11515 25220 Added a new server configuration parameter, tasker.uninterruptableSignalCascades, that controls whether slaves should process incoming job control requests when there is an existing signal cascade being processed. If TRUE, slaves will ignore incoming job control requests if an existing request is in-process and a message will be printed to the slave log indicating as such. Default is FALSE.
Accelerator VOV-11650 29856 Added an alert with error message to notify user when create reservation fails due to reservation period overlap with an existing reservation.
Accelerator VOV-12043 CS0120714 Calling vtk_slave_reserve will now cause a purge of all existing reservations for the given slave prior to new reservation creation.

Any vovslave or vovslavemgr start/restart, vtk_slave_define , etc invocation involving -e will also apply the same purge of all existing slave reservations before creation of a new reservation on the slave.

Users must use vtk_reservation_create to create multiple reservations on a slave, or via the web based UI.

Accelerator VOV-7591 20937 Documentation has been updated to not use RAM# in redundant quantitative resources example.
Accelerator VOV-4514 20957 Duplicate values in resource requests for consumables do not cause the job to be rejected in nc run.
Accelerator VOV-12003 CS0120875 Fix for cleaning up files under data/slaveload
Accelerator VOV-12095 AAP24206 Fixed "passing env variable to NC job inside container" issue.

User can call env FOO=BAR nc run ... and get correct value of $FOO inside the job.

Accelerator VOV-11892 CS0120652, CS0120732 Fixed a bug in vtk_flexlm_monitor when it would not create a summary resource map for a feature with multiple tags. The -order parameter to vtk_flexlm_monitor only orders tags, it no longer adds tags (use the -tags to add the tags).
Accelerator VOV-11014 24851 Fixed a timing issue where NC failed to properly reconnect to LM after an LM failover.
Accelerator VOV-11469 25174 Fixed an error that affected containerized jobs where at least one of the container hook scripts required root privileges. In this case, non-root privileged scripts as well as the job itself could potentially elevate themselves to root.
Accelerator VOV-11485 24206 Fixed an issue in the OS container support that could trigger an error message in the vovslave log about it being unable to "clear environment."
Accelerator, Accelerator Plus VOV-12200   Fixed issue with job arrays that are dependent upon a job or set, where the array jobs will not run due to an invalid input.
Accelerator VOV-11543 25205 Fixed issue with plotting workload wait reasons
Accelerator VOV-11699 29895 Fixed the issue where jobs are staying forever in pending state waiting for Limit:@USER@_N resource when both 'CGROUP' and 'limit' options are given in the job's input resource list.
Accelerator VOV-6077 21235 Fixed vovslavemgr restart issue when slave has jobs running.
Accelerator VOV-11895 AAP24898, CS0120786 Inequality comparisons using at-delimited field names in preemption rules (e.g. "PRIORITY<@PRIORITY@") will no longer give an error message about non-numeric comparisons.
Accelerator VOV-12125 CS0120945 Moved processing of interactive jobs (nc run -I, -Ir, etc.) to a separate thread for each job. This should improve the responsiveness of interactive jobs.
Accelerator VOV-11805 29905 Reported intermittent slowness in NC queue has been addressed in an earlier release.
Accelerator VOV-10356   Stderr will now be included in the output log for default windows jobs.
Accelerator VOV-11923   Stopping a job while using VOV_SLAVE_SID_DISABLE should no longer result in stopping processes in the same SID that are not part of the job.
Accelerator VOV-11298   The nc getfield command previously reported lowercase fields "curram", "slaveid", and "retracingid" with 0 values although the corresponding uppercase fields "CURRAM", "SLAVEID", and "RETRACINGID" were correctly reported. This fix corrects the reporting of the lowercase field values.
Accelerator VOV-12016 CS0120897 The duration of the reservation for the vovdbd vovslave is changed from 365d to 'forever', so that no regular jobs may be placed on it after expiration.
Accelerator VOV-8026 21721 The issue where vovslavemgr start needs to wait until vovfoster jobs has completed was addressed in an earlier release.
Accelerator VOV-6082 24694 This fix improves reliability of the -wl option to nc run by avoiding timing-sensitive collisions with other jobs on the ports used.
Accelerator VOV-12046 CS0120935 vovserver will manage the availability of resources on a slave without vovfoster jobs. When slave is requested to exit while it has a running job, it changes its state to DONE and renames itself to "name_stopped_id". Starting the same slave, at this moment, will result it to adjust the numbers of available resources based on their consumption in the previous instance.
Accelerator Plus VOV-11928 CS0120712 Added a safeguard mechanism to prevent a base queue job status query from blocking vovwxd operation. The mechanism is configurable by setting CONFIG(jobstat,timeout) in the SWD/vovwxd/config.tcl file to a valid timespec (eg 30s, 30m 30h) or number of seconds. The default timeout is 5m.
Accelerator Plus VOV-12220 CS0121046 Fixed a bug in vovwxd that was causing it to lose track of the slaves list and print redundant "Metrics for..." log messages
Accelerator Plus VOV-11715 AAP29909 Fixed an issue where an interactive job's (-I, -Ir, etc.) TERM environment variable would be incorrectly set to the user's TERM environment variable when the job's output is redirected (for e.g., piped). The job's TERM environment variable will now always be set to 'network' if the interactive job's output is redirected (for e.g., piped). This issue used to manifest itself only in jobs run by vovtaskerroot (not vovslave).
Accelerator Plus VOV-12267 AAP29909 Fixed an issue with interactive jobs (nc run -I/-Ir etc.) where the user is unable to interact with the job (including arrow keys, control key, etc. not working) when the output is piped to tee. Also, added an option (-forceterm) to disable setting the interactive jobs's TERM environment variable to "network" when the output is piped. This fix requires an update of vovsh and vovslave and a restart of any running vovslaves after the update.
Accelerator Plus VOV-11910 CS0120797 Fixed issue with WX slaves that prevented them from starting with a max life override of unlimited. Added log messages that denote when the override is activated during startup, and also the adjustment of unlimited max life to the greater of: a) the expected duration of the first job routed to the slave or b) the default slave max life as specified per the vovwxd configuration file.
Accelerator Plus VOV-11868 CS0120760 Improved the performance of checking the status of all pending ("requested") slaves when vovwxd is querying the status from the base queue(s).
Accelerator Plus VOV-11236 24875 Stack trace generation reworked to provide correct symbols and core generation thus providing the correct source of a fatal error.
Allocator VOV-11353 25000 Better handling of auxiliary save proc cleanup.
Allocator VOV-11625 AAP25280 Made LA behaviour of marking vendor-queue tokens as out-of-queue tokens configurable using the config settings 'LA::MarkVqAsOoq' and 'LA::MarkResourceVqAsOoq' in vovlad/config.tcl. The default behaviour is same as existing behaviour.

'LA::MarkVqAsOoq' can be used to modify LA-wide behaviour while 'LA::MarkResourceVqAsOoq' can be used to modify the behaviour for a specific given resource.

FlowTracer VOV-11835 AP29893 Fixed an issue where indirect slaves were consuming seat licenses under FlowTracer.
FlowTracer VOV-11834 CS0120713 Fixed an issue which resulted in an indirect slave being stopped or hung when a job running on it was stopped.
FlowTracer VOV-11874   Fixed an issue with FlowTracer and vovwxd on LSF where alerts of the form "Number of allowed slaves has reached limit... Consider increasing maxNormalClients in WX server, or max. slaves in vovwxd config..." would be continually generated.
FlowTracer VOV-10934 24730 Fixed the issue with custom color set in gui.tcl using vtk_transition_color_control are not getting reflected in vovconsole when node status changes to Failed or Valid.
FlowTracer VOV-11920   Fixed the issue with custom node color not getting reflected under set view when the color is set using vtk_nodestatus_set_color in gui.tcl
FlowTracer VOV-11890 CS0120807, CS0120788 If the vovserver receives a SIGTERM signal it will perform a graceful shutdown in a manner similar to autostop.
FlowTracer VOV-12420   In vovwxd, the user can set CONFIG(dequeueExtraSlavesEnable) 0 parameter in config.tcl which will stop the creation of "bkill" jobs for any pending slaves.

The user can also undefine the GetJobKillCmd procedure from vovlsf.tcl for the same purpose.

FlowTracer VOV-11873   To enable resources to be managed locally by FlowTracer when using an LSF backend via vovwxd enable the vovwxd.localresources parameter in the policy file and create the resources in FlowTracer using the vtk_resourcemap_set -local parameter.
FlowTracer VOV-11921   vovconsole behavior has been modified for stop/invalidate menu options so that all selected jobs and down-cone along with the jobs in nested sets will be stopped/invalidated hierarchically.
FlowTracer VOV-11837   vovwxd no longer uses reservations for resources. Instead, to enable resources to be managed locally by FlowTracer when using an LSF backend via vovwxd enable the vovwxd.localresources parameter in the policy file and create the resources in FlowTracer using the newly added vtk_resourcemap_set -local option.
FlowTracer VOV-11869 CS0120761 vovwxd should be used instead of vovlsfd. vovwxd no longer uses reservations. Instead, to enable resources to be managed locally by FlowTracer when using an LSF backend via vovwxd, enable the vovwxd.localresources parameter in the policy file and create the resources in FlowTracer using the vtk_resourcemap_set -local parameter.
Monitor VOV-7501 20772 Duplicate of VOV-7082, VOV-9346 tracked by CSG cases 20029, 20772, 23535, 24288
Monitor VOV-12068 CS0121142 Fixed an issue on Windows-based platforms where accessing CGI-based web UI pages via the read-only guest access port resulted in the administrative privilege for the product instances instead of the intended read-only privilege.
Monitor VOV-12214   loadremotedata sub-command for lmmgr had mistakenly been disabled. The sub-command has been restored.
Monitor VOV-11586 25124 Fixed an issue where configuring Altair Monitor to monitor its own usage of a keyfile license was not working correctly.
Monitor VOV-9346 23535, 24288 SafeNet Sentinel HASP monitors now support the latest Admin API version.