2023.1.0 Release Notes
New Features
Internal Number | Products | Case Number | Description |
---|---|---|---|
VOV-3411 | Accelerator | None | vovselect TURBOBOOST from taskers is now available. Set to 1 if Intel Turbo
Boost is enabled on this tasker host, or 0 if disabled (Linux
only), otherwise set to NA. |
VOV-7894 | Accelerator | AAP21491 | Added Jobclass and Autokill columns to the Running Jobs table in the web UI. |
VOV-8024 | Accelerator | AAP21343 | Added the ability to specify a stop reason when stopping all running jobs for all users via the Running Jobs page. |
VOV-8642 | Monitor | None | The ftlm_tag_admin command line utility for Monitor is now described in the documentation. |
VOV-9900 | None | AAP24167 | An enhancement has been made for the NC family of managers (ncmgr, vncmgr, wxmgr and hemgr) to show the queue name and command action in the log file name if logging to the $VOVDIR/local/logs/* location. |
VOV-12011 | Accelerator | CS0120886 | A new ncmgr rehost" subcommand is added to migrate an Accelerator vovserver from one host to another. See ncmgr rehost for details. |
VOV-14032 | All | CS0223251 | vncmgr/ncmgr stop action now supports a -freeze_nocpr option to save the PR file without compression for a freeze to potentially save time when working with very large PR files. |
VOV-14383 | Accelerator Plus | None | The Direct Drive documentation topic has been clarified with more information regarding formatting of the driver script. |
VOV-14395 | Accelerator | None | The vtk_server_config Tcl command has a new rds.reinitialize parameter to allow reinitialization of the RDS system from the command line. This parameter requires no additional arguments. |
VOV-14488 | Accelerator Plus | CS0260943 | Added new configuration parameters for fine-grained control of vovwxd. |
VOV-14514 | Accelerator, Accelerator Plus | CS0268091 | Clients that interact with VOV subsystems are "immediately" terminated if they do not have sufficient security privileges (as defined in security.tcl). The ANYBODY / READONLY security level is documented for users with minimum privilege. |
VOV-14577 | All | None | The default HTTP server, or webprovider, used by vovserver is changing to the internal webprovider in place of "nginx". This applies to the URL used to access the vovserver "web port". The -webprovider nginx option is available on ncmgr and other vovserver start commands to select the nginx webprovider if that is preferred. |
VOV-14701 | Accelerator | CS0292715 | You can now use vovselect to list tasker lists created with vovtaskerlist, and get information about tasker lists via the v3 REST API with the URL http::hostname/api/v3/taskerlists. |
VOV-14718 | Accelerator | None | Implemented backfilling resource reservation to handle conflicting jobs. This can be controlled by server config backfillResReservation, disabled by default. |
VOV-14774 | Accelerator, Monitor | None | Four new subcommands are added to the vovdb_util command for managing the Monitor or Accelerator PostgreSQL database: exportconfig, exportpasswords, importconfig, and importpasswords. See vovdb_util for details. |
VOV-14789 | Accelerator | AAP22106 | The use of access control lists (ACLs) for resource maps are expanded to allow specified
users or user groups to edit, reserve, and forget (delete) the
resource map object. An example of an ACL-setting command that
would be used for this objective is:
where
$id is the VOV id number for a resourcemap
object. Supported only if the resourcemap is created with
rank=0. |
VOV-14806 | Accelerator | None | Rapid Scaling 3.0 implemented in vovwxd and the Altair NavOps cloud connector. Supports in-cloud deployments of a cluster scheduled by the Accelerator workload manager with adaptive compute node allocation to adjust to job load. |
VOV-14863 | Accelerator Plus, Monitor | CS0312583 | Enabled the doTestHealthFailoverServerCandidates health check for Accelerator Plus and Monitor. |
VOV-15120 | Accelerator | None | Implemented backfilling resource reservation with non-conflicting jobs. This can be controlled by server config backfillResReservation, disabled by default. |
VOV-15352 | Accelerator | None | Implemented dispatch of deadline jobs on lookahead reservation activation. Supported argument -deadline in nc run to submit the job with an expected duration of completion. |
VOV-15404 | Monitor | None | Migrate replacing images from Monitor batch reports to node, only D3 based plots are now supported which are part of the new look and feel of Monitor. |
VOV-15429 | All | None | The default license manager for Accelerator products is changed to Altair License Manager in
the 2023.1.0 release. To revert to RLM, set
alm.enable to 0 in
policy.tcl. This change does not affect
licensing by the RTDA legacy keyfile license files, which are
still supported. |
VOV-15439 | Accelerator | None | Implemented tasker property hardbound and softbound to
restrict the scheduler to dispatch only autokill jobs on
hardbound tasker, while autokill & xdur jobs may run on
softbound tasker. Implemented reservation property hardfill and softfill to restrict the scheduler to backfill hardfill reservation with only autokill jobs, while softfill reservation may be backfilled with autokill and xdur jobs. Modified vovtaskermgr utility to support all 4 options. |
VOV-15548 | Allocator | None | Implemented Allocator Resource Summary page column 'Allocated' to show the "Number Of Resource Tokens Allocated Including OOQ Tokens To Each Site". |
VOV-15588 | Accelerator | None | An online help topic to describe AVS syntax has been added. |
VOV-15618 | Accelerator | None | DP component jobs now inherit the expected duration (-xdur) setting of the main job. |
VOV-15653 | Monitor | None | The PostgreSQL software included with Monitor and Accelerator products is upgraded to version 14.4. A database conversion to the new version is needed for projects that are upgrading and retaining an existing database. See Monitor and software installation documentation for database upgrade instructions. |
VOV-15656 | Accelerator | None | The "tasker" lexicon versions of Accelerator CLI commands were added in a recent software
release to replace the "slave" lexicon versions. The "slave"
lecicon commands have been available, but be advised of their
deprecation. The following CLI commands are deprecated and will
be removed in an upcoming release:
|
VOV-15661 | Accelerator | None | The following new server configuration parameters are added to control the new lookahead and
backfill scheduling policies. See
policy.tcl for descriptions:
|
VOV-15694 | Accelerator | None | Resource Data Service (RDS) is a replacement for vovresourced. RDS is provided as an opt-in preview feature. When the optional RDS operation mode is selected, Accelerator and Monitor communicate on new pub/sub channels. Resource management and license matching is performed by a new RDS thread of the vovserver process. |
Resolved Issues
Internal Number | Products | Case Number | Description |
---|---|---|---|
VOV-11585 | Accelerator Plus | AAP25265 | Fixed a bug in vovwxd that caused incorrect "taskers max reached" alert. |
VOV-12137 | Accelerator | CS0120974 | nc modify now accepts the syntax to change the FairShare group of a queued job, and will allow the user to be different than the job's original FairShare user so long as the client has the correct ATTACH ACL on the new fsgroup. |
VOV-12766 | FlowTracer | None | Changes in functionality of vovforget. In Accelerator, Accelerator
Plus, Hero:
|
VOV-13525 | Accelerator, Accelerator Plus | CS0288823 | Fixed an inconsistency in tasker status returned by nc hosts and vovselect. |
VOV-13784 | Monitor | CS0191792 | The client systems may have a clock skew. Therefore timestamp reported by client should not be used for calculations. For consistency, from now on, we avoid client reported timestamps for job checkouts, checkins, matching, etc. With this fix, server-side timestamp are now prioritized. |
VOV-14321 | Accelerator | CS0246573 | Resource requests attempted with #0 or /0 will now throw an error through out the CLI,API, and UI, with the exception of CORES/0, or CPUS/0. |
VOV-14356 | Monitor | CS0247184 | The display of incorrect status of LM Control Center Agents in Web UI has been fixed. |
VOV-14501 | All | CS0262252 | vovserver config parameter autoLogout now allows admins
to configure the maximum duration for which a given Web/REST
session stays valid. |
VOV-14505 | Accelerator, Accelerator Plus | CS0263183 | The bjobs implementation has been improved to significantly reduce the number of vovserver inquiries. |
VOV-14628 | Allocator | CS0270617 | A tuning parameter was added to the LA distribution algorithm. It can be modifed using the
command LA::SetUnhappyReserveFraction
<fraction> in <swd/vovlad/config.tcl. The
fraction must be a number between zero and one and defaults to
zero. A value of zero corresponds to the behavior before this
change as implemented. |
VOV-14824 | All | None | Fix a denial-of-service (DOS) vulnerability in vovserver's base webserver from nefarious large HTTP requests. |
VOV-14886 | Accelerator Plus | CS0266302 | Added support for a new server configuration param called wx.setQueueEnv
to control how NC_QUEUE may be set in the job environment. If not set, NC_QUEUE will be set on the job environment only if it's explicitly set in the job submission environment and the user is using SNAPPROP or SNAPSHOT. If wx.setQueueEnv is set to 1 and NC_QUEUE is not set in the submission environment, NC_QUEUE in the job environment will be the WX project name. If NC_QUEUE is set in the submission environment and SNAPPROP or SNAPSHOT is used, NC_QUEUE will be propagated from the submission environment to the job environment. |
VOV-14959 | Accelerator Plus | CS0267206 | Fixed a bug in vovwxd that caused an infinite loop due to a deleted "top job" of a bucket. |
VOV-14985 | FlowTracer | CS0324179 | Introduced vtk_set_operation MOVECONTENT to transfer content from one set to another. |
VOV-14990 | Accelerator Plus | CS0336870 | Fixed a bug causing a vovDeprecated TCL error in vovwxd. |
VOV-15105 | FlowTracer | 00000 | The output files of autoflow jobs are no longer marked missing by vovcheckfiles. Moreover, direct descendents of autoflow jobs are always considered valid provided the parent node is valid. |
VOV-15112 | All | None | Updated Kafka Job events with additional data. |
VOV-15188 | Monitor | None | The code management system used to manage license files has changed from CVS to fossil. New installations will use the fossil based system by default, existing installations will continue to use the CVS based system by default. If the user wishes to switch from the CVS based system to the CVS based system, some manual steps will be required. The user can choose start afresh and delete the existing CVS history, or choose from two recipes to migrate the CVS history to the fossil based system. |
VOV-15195 | FlowTracer | None | Fixed job name issue with IFDEF FDL procedure that prevented it from running successfully. |
VOV-15257 | All | None | OpenSSL version on Windows and Linux platforms is upgraded to 1.1.1p |
VOV-15351 | FlowTracer | CS0352183 | vovwxd added support for the CONFIG(slave,env) parameter in DirectDrive mode. |
VOV-15387 | FlowTracer | CS0360469 | The default values for VOV_SEND_TIMEOUT_MS and VOV_RECV_TIMEOUT_MS on Linux will now be set higher to 2 minutes connection timeout and 5 minutes for send/receive timeouts. This should mitigate protocolErrorInBuf type errors, with the default values under most conditions. If higher or lower values are desired the user can still override the default behaviors by setting VOV_SEND_TIMEOUT_MS and VOV_RECV_TIMEOUT_MS as desired for any particular scenario. |
VOV-15416 | FlowTracer | CS0362761 | The vovproject archive command no longer truncates the hostname. |
VOV-15417 | Monitor | None | Fixed issue that prevented the lmmgr reset operation from completing without error. |
VOV-15482 | Monitor | CS0332038 | Monitor's batch reports are using the same version charts like the ones found in the browser setting by default. The use of the command line utility remains the same. However, in order to extract the charts and convert them to static images new requirements have been set in place. Node v14 or greater is required to have been installed on the machine and be available in the path environment variable. In addition, manually reverting the LM configuration parameter "lookAndFeel" to its previous value will result to image extraction failure. |
VOV-15498 | Monitor | None | Remove the "usage plot" and "plot queued results" checkboxes. The functionality of these checkboxes is replaced by: "plot usage details checkbox" and "plot usage capacity" checkbox for the usage plot and "plot queued request details" for the queued plot. The default behavior of the form remains the same. Visiting the page and submitting the form without editing any of the options will result to the exact same plots. New checkbox with the label of "plot average queued" that will enable the plotting of the average queued requests, thus decoupling the "show average" checkbox from controlling both plots. Any component/element of the two plots can be plotted irrespectively of the selection made in the checkboxes. For example for the usage plot, the average can be set to on without the need to have the details or the capacity on the plot. The checkboxes have been re-ordered and groupped to accordingly to the plot they have control over. |
VOV-15515 | Accelerator | CS0373997 | Fixed a bug in tracking of resource usage by taskers that are in "DONE" state. |
VOV-15541 | Accelerator Plus | CS0376655 | Using containers could cause some potential naming conflicts in scripts when based upon timestamp and pid alone, which unfortunately is not guaranteed to be a unique identifier with current configuration options for containers. There a true uuid has been added in utility scripts for commonly used scripts where such ambiguity might normally cause issues when used frequently inside multiple containers at the same time. |
VOV-15560 | Allocator | None | Fixed issue in Allocator resource summary page which is resulting in zero size csv file. |
VOV-15620 | Accelerator Plus | None | A -dd option was added to vovwxconnect that uses Direct Drive when connecting to NC queues. The -dd option can also be passed to wxmgr start to connect to existing queues, for example, wxmgr start -basequeue vnc -dd. |
VOV-15724 | Accelerator, Accelerator Plus | CS0398311 | Fixed a bug causing a slowness in processing foreign buckets in WX DirectDrive mode. |
VOV-15735 | Accelerator, Accelerator Plus | None | Fixed issue that resulted in permanently-used resource maps when the top job of a foreign bucket can not be created. The issue commonly showed up in NC base queues with a WX meta-scheduler using direct drive. |
VOV-15741 | Accelerator | None | Reject Accelerator jobs that consume no slots. The fix detects this case: nc
run -r slots/0 ... and rejects the attempted job
submission. |