2016.09 Update 9

New Features and Enhancements

Products Internal Number Case Number Description
All 7697 21191 In Admin/Database page or vovdbd_util, user is able to specify port for use by the VOV Postgres DB.
All 7657 The slave will now print a warning message to its log when the server appears to be a different version from the slave.
All 7728 Allow colons in property names.
FlowTracer 7788 Added option to vovbuild to prevent building when jobs are running.
FlowTracer 7743 21229 When vov/vw/vrt is used to launch jobs from a client, and an illegal option is passed, make sure to update the "WHY" property to tell the user exactly why it failed. Previously it would exit with error code 2 and not provide any reason why.
FlowTracer 7609 21025 Display run and suspend times in UIs that show job duration.
FlowTracer 6852 15427 In FlowTracer and NC GUI, added subsets to the list of nodes that "quick find" highlights and selects. Previously it would only search for jobs and files.
FlowTracer 7667 In FlowTracer and NC GUI, the quick find search field searches every time you type or click the mouse in the search field, provided there are at least 2 characters. For large node counts, this made it feel like it was hanging between keystrokes. In this fix, it initiates a search request when the search text is changed, but does not start the search until a small amount of time passes, so it prevents starting the search too early. It also can interrupt the previous search if more characters are typed. In addition, a bug was found that was causing it to search for each node type twice. That bug was fixed. For large node counts (100k+), it prints white status messages on the screen to show the search progress. The result is a faster and more responsive quick find feature.
FlowTracer 7689 When NFS protection is enabled using the nfsDelay setting in policy.tcl, do not wait for the nfs delay time to pass when checking if a file exists if the file's status is MISSING.
FlowTracer 7688 When using vovbuild, if an input or output dependency is excluded due to exclusion rules such as those in the server working directory file "exclude.tcl", print a warning message.
LicenseAllocator 7105 20054 Site nicknames will no longer be truncated in the column headers of the overview page - the full nickname will be displayed.
LicenseAllocator 7576 20897 LA overview page will now show the sum of all running jobs for each resource, and all queued jobs for each resource, across all sites.
LicenseAllocator 7581 20901 Clicking running or queued jobs in Overview page or Resource Summary page will take you to an NC page that only shows jobs that are running with or queued for this specific resource.
LicenseAllocator 7279 20452 The resourcemap owner name will now include the name of the LA project which is currently controlling this resource.
LicenseMonitor 7706 Added CLI version of the LM agent SFD for Windows. The new agent, lmagent-win64-cli.exe, can interact with the Windows SCM (requires UAC elevation). Run with -h to get the full usage syntax.
NetworkComputer 6022 21107 Added option -q to nc forget. Note that the command is not quiet in case of errors. To make it completely silent, use: nc forget ... >& /dev/null.
NetworkComputer 7452 20720 The vovreconciled daemon now includes the name of the license when notifying that reconciliation is skipped for a job.
NetworkComputer 7652 21013 Added new nc info -dep command to easily list job dependencies: 
> nc info -dep 7299670
 007299670 dependees:	007299667 
 007299670 dependents:	007299673 007299676 007299688
HERO 7103 20071 Completed implementation of vtk_fs_stat by supporting the field "fsid".

Resolved Issues

Products Internal Number Case Number Description
All 6966 If a server log file cannot be opened at server startup, the server will exit with a fatal error. If a new server log file cannot be opened during daily rotation and the previous log file is not accessible, a new log file will be opened under tmp directory.
All 7381 20552 Added a column to show resource reservations in the monitor GUI.
All 6921 Slaves will timeout predictably, after attempting to connect/reconnect to server. VOV_RELIABLE_TIMEOUT can be used to specify the timeout period. Command line option -t can be used to override the setting of VOV_RELIABLE_TIMEOUT.
All 7524 20809 Now db host can be configured using fqcn, alias, or IP address, in addition to the canonical short host name. If host provided cannot be pinged, an error message will be displayed.
All 7809 21345 Fixed issue that caused the patch status link in certain web UI pages to be rendered as plain text instead of HTML.
All 7698 vovconsole -geometry option was ignored in 2016.09 u8. Fixed to make -geometry option respected.
All 7801 No longer show utilization percentage in monitor GUI for resources with unlimited capacity.
FlowTracer 7777 21297 Added option to vovbuild to prevent building when jobs in the flow are running. The new option is vovbuild -c which is short for "cautious" mode.
FlowTracer 5423 21190 vovlsfd will generate a warning if it detects jobs with expected duration longer than MaxLife parameter. Slaves will not be submitted for these jobs.
FlowTracer 7814 Fixed a bug that caused printing an incorrect count of the number of overloaded slaves in the output of vsy, nc why, wx why and vtk_explain_status. The formatting of the output of the vsy, nc why, wx why, and vtk_explain_job_status commands has changed in the following ways:
  • In the "Server Summary" section, we now print "Hardware conflicts found" instead of "No compatible slaves found".
  • In the "Server Summary" section, we now print "Job cannot be run because the following resources are not currently free on any slave" instead of "Job cannot be run because the following resources do not currently exist on any slave".
  • In the "Slave Summary" section, we now print "Final available candidate count: 0" in the event of hardware conflicts that prevent any slave from starting a job.
  • In the "Slave Summary" section, we now print "(ignoring load)" instead of a meaningless overloaded count when -i is specified for vtk_explain_job_status.
  • In the "Additional Information" section, the status "Full" is now written as "Fully Loaded".
  • In the "Additional Information" section, slave statuses are separated from slave counts by a colon (:) character.
FlowTracer 7694 Sending a SIGTSTP signal to an indirect slave no longer kills the slave. Instead the slave process and all of its jobs will be suspended until another signal is received.
FlowTracer 7707 Fixed vtk_server_config protocol error in readonly security level.
FlowTracer 7724 Redirected indirect slave log from stdout to the slave daily log file.
FlowTracer 7725 Fixed grid mode drawing bug that made FlowTracer seem to go into an infinite loop of drawing when the setting to move nodes to the end in grid mode was enabled and the graph was already maxed out (autofit overflowed the bottom due to too many nodes to fit in the canvas).
FlowTracer 7590 20930 Corrected vovps documentation, and added example.
LicenseAllocator 7222 21220 LA now correctly drops overbooking to zero when overbooking is turned off.
LicenseAllocator 7794 21306 LA will try to meet demand first before distributing extra tokens. Also, LA will not take tokens away from critical sites when adjusting for uncertains.
LicenseAllocator 7796 21311 LA will use wave based OOQ values for computing available tokens for distribution.
LicenseMonitor 7647 20906 Fixed issue with loading remotely-generated data via the lmmgr loadremotedata command caused by the utility not connecting to the LicenseMonitor server.
LicenseMonitor 7676 21269 On Windows, one of file path modules tried a shortcut expansion on a regular file and reported that the file doesn't exist even though the file exists. Fixed to try the shortcut file expansion when it is a really a shortcut. Also added recovery codes when the expansion fails.
LicenseMonitor 7665 20884 Fixed an issue in the detailed plot report when showing reservations that caused currently-active reservations to be displayed as normal checkouts in the plot.
LicenseMonitor 7704 21178 Fixed issue with batch reporting of denial plot for multiple features that resulted in the carryover of previous plot data for subsequent features that have no denials.
NetworkComputer 5624 20532 Changed behavior for when we reconcile a resource R we can take a look at the previous resource in the grabbed list and check if the previous resource is a summary resource for R.
NetworkComputer 7680 Fixed bug where exporting preemption rules previously wrote out incomplete rules in some situations.
NetworkComputer 7699 20879 Fixed bug where removing resources from slaveClass.table would not remove them from the server without a restart.
NetworkComputer 7770 20883 Output files are no longer tracked for jobs submitted via the LSF emulation layer. This prevents a queue bucket from being created for each job that uses the bsub -o option to specify the job log file.
NetworkComputer 7607 20982 LSF bsub emulator uses VOV_JOBPROJ environment variable for the project name. Specifying a project explicitly with -P in bsub command overrides the project specification in the VOV_JOBPROJ variable.
NetworkComputer 7636 20603 Fixed a bug that caused VOV_PORT_NUMBER not to be applied correctly after a connection timeout. Previously, only the last port listed would be used after VOV_RELIABLE_TIMEOUT had expired.
NetworkComputer 7641 Preempting and preemptable selection rules are now validated at preemption daemon start time. Preemptable rules are also validated upon each update of the rule as job queue buckets change.
NetworkComputer 7682 21153 Fixed behavior so that all NC jobclasses from $VOVDIR/local/jobclass show up in the web GUI.
NetworkComputer 7708 Fixed issue with slave-reservation preemption that treated this HW-based rule as one that is SW-based, preventing all related preemption attempts from succeeding.
NetworkComputer 7822 21365 Corrected the vovslavemgr reserveshow output format for unending reservations.
NetworkComputer 6498 NC and WX now use the 'any' port mode by default, to help prevent deadlock situations when the port chosen by the previous default of 'automatic' mode is already in-use.
NetworkComputer 7588 20923 Fixed issue that prevented auto-rescheduled jobs from avoiding the previously used host. Also added a new configuration parameter in the policy.tcl file, autoRescheduleOnNewHost, to control whether autorescheduled jobs will avoid an entire host, or just the same slave. Set to 1 (default) for entire host, set to 0 for same slave.