processes (Process Monitoring) Release Notes

Contents
uimpga-ga
processes_RN
The (Process Monitoring) processes probe monitors the specified processes to detect any error situation. The probe also retrieves information about the process, for example, the CPU usage, memory usage, and so on. The probe verifies the following conditions:
  • A process is expected to run, but has stopped.
  • A process is not expected to run, but is running.
  • A process does not run as expected.
  • The number of process instances is not as expected.
  • The CPU usage for a process is above threshold.
  • The handle counts for a process are above threshold (Windows only).
  • The memory usage for a process is above threshold.
  • The thread usage for a process is above threshold.
Contents
Revision History
This section describes the history of the revisions for this probe.
Support case(s) may not be viewable to all customers.
Version
Description
State
Date
4.70
What's New:
  • Stabilization of the processes probe for Memory consumption and CPU Usage.
  • VS2017 redist package version upgraded to 1.01 from 1.00 as 1.00 was having reboot issue.
Users may see some CPU spike while opening the IM processes GUI or while refreshing the processes list in GUI. This CPU spike will go down, once the GUI operation gets closed.
Please refer Upgrade Considerations for the upgrade.
Fixed Defect:
  • Processes Probe - 4.64 CPU leak on CentOS 6.10 and Ubuntu 14.
    Support case number
    20020534
  • Deployed processes probe version 4.64 to AIX systems and caused issues.
    Support case number
    20024732
  • Processes probe error: Internal error: Unable to find any processes.
    Support case number 20049414
  • Process probe on AIX 7.1 consuming high cpu.
    Support case number 01372650
  • Processes probe memory and CPU usage is increasing.
    Support case number 01301068
  • Problems with the probe processes.
    Support case number 01344072
  • Processes probe creating cpu spikes of around 5% when polling on zlinux.
    Support case number 20101877
  • Linux processes probe - memory leak.
    Support case number 01332245
  • UIM processes probe dependency is for vs2017_vcredist_x version 1.00.
    Support case number 20092103
GA
February 2020
4.64
What's New:
  • (May 2019)
    Updated the processes MCS (version 4.65 onwards) as part of addressing the common vulnerabilities and exposures by updating the jackson-databind libraries. For more information and CVE numbers, see Addressing Jackson Vulnerabilities.
  • Added support to monitor metrics on per process basis to determine the process that causes heavy I/O.
  • Added support to monitor the number of files open for processes that impacts the Unix and Linux servers performance.
GA
February 2019
4.63
What's New:
  • Updated this probe as part of removing dependency on the end-of-life (EOL) Microsoft Visual C++ Redistributables in CA UIM 9.0.2. CA UIM 9.0.2 now uses Microsoft Visual C++ Redistributable for Visual Studio 2017.
  • Updated this probe as part of removing known security vulnerabilities in CA UIM 9.0.2 by using the upgraded OpenSSL components.
  • (December 2018) Added support for monitoring Amazon Linux 2.
  • (February 2019) Added support for monitoring Windows 2019.
GA
October 2018
4.60
What's New:
  • Added support to monitor each instance of a process where the process is recognized by its name and process ID.
  • (June 2018) Added support for monitoring Windows 10 (x64).
  • (June 2018) New MCS template (processes_mcs_template 4.61). Enhanced profiles that enable you to configure metrics, baselines, alarm thresholds, alarms - including Time Over Threshold alarms - and custom alarm and close alarm messages, all within a single MCS profile. For more information, see Configuring Alarm Thresholds in MCS.
  • (July 2018) Added support for monitoring RHEL 7.4 x86 and x86 64-bit, and zLinux 64-bit.
  • (September 2018) Added support for monitoring Ubuntu 18.04.  Apply robot 7.93HF10 for this support.
    Support Case: 01184423
Fixed Defect:
  • The QoS interval time was slightly delayed for the processes state monitor.
    Support case number 781481
GA
September 2017
4.50
What's New:
  • (IM Interface only)
    Added support for:
    • AES-128 encryption
    • Monitor Windows systems where FIPS encryption is enabled. Before you deploy the probe version 4.50, you must enable the FIPS encryption on the system where the probe is configured.
      You cannot configure the probe from any Admin Console (AC) interface on a FIPS encryption enabled environment.
CR
May 2017
4.40
What's New:
  • (Windows)
    Added support to monitor the symbolic links of a process using binary path.
  • (Windows)
    Added the
    useBackslashInProcOwner
    key that allows you to use "\" as the domain name separator while setting the process owner for a process.
  • (For HP-UX only) Added the
    get_full_command_line
    key that allows you to view full command line parameters for a long process. For more information, see View Full Command Line Parameters.
    Support case number 632745
Fixed Defects:
  • The probe calculated incorrect QoS data.
    Support case number 472252
  • The probe generated lots of errors in the probe logs that resulted in flooding of the log file.
    Support case number 676172
GA
April 2017
4.32
What's New:
  • The probe supports monitoring within an IPv6 environment.
Fixed Defect:
  • The probe did not clear pending alarms in the following situations.
    Support case number 431592
    • Any profile deleted or deactivated
    • Any profile modified and saved without hard restart of the probe
    • Probe upgraded with pending alarms of deactivated profiles. If you have already upgraded the probe, edit the profile and save the configuration.
GA
September 2016
4.31
What's New:
  • (Solaris and HP-UX)
    Added support to monitor the symbolic links of a process using binary path.
  • Added support to monitor binary paths of processes with command arguments.
  • Added support to expand the
    $process
    variable in
    MsgProcessInstances
    alarm with the binary path of processes.
GA
May 2016
4.20
What's New:
  • (Linux and AIX only)
    Added support to monitor the symbolic links of the process using binary path.
Fixed Defect:
  • On restarting, the probe removed the clone profiles from the Profiles list.
    Support case number 347255
GA
May 2016
4.10
What's New:
  • (Linux and Solaris only)
    Added support to monitor the resident memory usage of the processes.
Fixed Defect:
  • The probe did not clear the pending alarms of the modified profile on soft restart.
    Support case number
    312559
  • The probe did not start on Windows due to a dependency on vs2010 redistributable.
    Support case number
    329119
    The probe now requires vs2008.
April 2016
4.02
What's New:
  • (For Linux only) Added the stack_size_in_mb key that allows you to search for processes using complex regular expressions. For more information, see Increase Stack Size.
    Support case number 246754
Fixed Defects:
  • The probe did not clear the "wrong process owner" alarm when process was started with the correct owner name.
    Note:
    The probe does not generate the clear alarm if the threshold condition is no longer breached or probe is restarted for reason such as configuration changes.
    Support case number 245804
  • The probe crashed during a scheduled task and performed core dump.
    Support case number 70000540
  • The probe did not send alarms with the updated alarm severity.
    Support case number 265085
  • (For Linux only) The probe crashed when complex regular expression was used to recognize a process.
    Support case number 246754
  • Updated the documentation to include the scenarios under which the alarm is generated during process up, down, or restart. For more information, see Admin Console (AC) configuration or Infrastructure Manager (IM) configuration articles.
    Support case number 246157
  • Updated the documentation to include the steps to stop truncation of more than 80 characters for Java process in Solaris platform. For more information, see Remove Java Processes Character Truncation.
    Support case number 70001043
February 2016
4.01
Fixed Defects:
  • The probe was not using customized clear message text in alarms.
    Salesforce case 00163151
  • The probe was unable to monitor processes and generate alarms if the used memory was greater than 100 MB.
    Salesforce case 00162654
  • When the probe reported a process as down, it did not generate process down alarm  with modified message text.
    Salesforce cases: 00161389, 00160183
  • Processes profile script was not executing properly.
    Salesforce case 00158405
  • On Unix platform, the QOS_Definition for the following IPC counters sent an incorrect value for the hasmax field.
    Salesforce case: 00167456
    • QOS_IPC_SHARED_MEMORY_SEGMENTS_UTILIZATION
    • QOS_IPC_MESSAGE_QUEUES_UTLIZATION
    • QOS_IPC_SEMAPHORE_SETS_UTILIZATION
    • QOS_IPC_PROCESS_UTILIZATION
GA
July 2015
4.00
What's New
:
  • Upgraded OpenSSL to version 1.0.0m.
June 2015
3.92
Fixed Defect:
  • All the eight IPC Counters were applying to the template on the Windows OS instead of applying the following supported templates:
    • Number of Processes
    • Number of Semaphore Sets
March 2015
3.91
What's New
:
Added IPC Counter monitoring for  Message Queues, Semaphores, Shared Memory Segments for Linux, Solaris, AIX, and Windows platforms.
Fixed Defects:
  • Fixed a defect where the probe did not clear the process up and down alarms.
    Salesforce cases: 00156008, 00156866, 00157484, 00154035, 00156940, 00156992, 0015651
    0
    Note:
    While upgrading the probe version from 3.83 to 3.91, process up and down alarms do not clear from Alarm Console automatically. User has to acknowledge these alarms manually.
  • Fixed a defect where the probe did not clear alarm from Alarm Console for CPU usage.
    Salesforce cases: 00155806, 00153144
March 2015
3.83
Fixed Defect:
  • The probe version 3.81 did not clear the process up or down alarms on the Alarm Console.
    Salesforce case 00146533
February 2015
3.82
Fixed Defects:
  • The probe generated only one process_down alarm for all closed instances of a process.
    Salesforce case
    00135021
  • The suppression key was missing for process restart alarm.
    Salesforce case
    00143378
  • The probe version 3.81 did not clear the process up or down alarms on the Alarm Console, when the process was stopped.
    Salesforce case
    00146533
  • The probe version 3.81 did not clear the Expected Instances alarms, when the process returns to the expected instance number.
    Salesforce case
    00148604
  • The probe did not clear the alarms when the process returns to the expected instance number.
    Salesforce case
    00151678
December 2014
3.81
Fixed an issue where certain alarms were being suppressed on selecting the Track Processes by Process Identifier checkbox, while creating a profile.
Salesforce case
00144501
October 2014
3.80
Added support for zLinux Operating System.
June 2014
3.77
Fixed Defects:
  • The probe did not delete a message from the message pool even when the user tries to delete the message without clicking Apply/OK.
  • The probe generated alarm with the default text and not the modified text even though you have edited the default MsgProcessRestart message, and restarted the probe and the process.
April 2014
3.76
Fixed Defects:
  • User was not able to add custom messages with clear severity in the Message Override list.
  • User was not able to get alerts with correct username (UID was getting displayed instead of username) when the username characters are more than 8.
February 2014
3.75
Fixed issue regarding probe defaults.
November 2013
3.74
  • Updated the probe defaults.
  • Fixed memory leak issue on AIX.
  • Fixed an issue related to process name which starts with square bracket.
November 2013
3.71
  • Fixed: GUI bug in processes related to exclude schedule.
  • SOC Defects Fixed
January 2013
3.70
  • Functionality to use user/ruser in ps command for UNIX added.
  • Added probe defaults.
  • Fixed incorrect metric definitions.
June 2012
3.63
Fixed an issue for process stop action alarm related to expanding $errmsg variable.
June 2011
3.62
Fixed hang situation on individual process monitoring.
June 2011
3.61
Fixed High individual process CPU usage.
June 2011
3.60
  • Fixed incorrect CPU utilization issue.
  • Added support process handle count monitoring in Windows environment deployments. This feature is not applicable for non-windows platforms.
  • Added support for clearing alarms on restart for the profiles that are no more in the alarm state.
  • Fixed an issue related to $expected_user alarm variable expansion.
  • Added support for monitoring thread count for processes on AIX, LINUX & SOLARIS.
  • Added support for overriding QoS target at profile level.
  • Fixed a crash in thread count monitoring.
  • Fixed Service Oriented Configuration defect.
  • More error checking when fetching performance data.
May 2011
3.52
  • Added fixes for web-based Service Oriented Configuration.
  • Fixed CFX File.
January 2011
3.51
  • Added option to limit processes that are seen to a configurable zone (Solaris).
  • Added support for internationalization.
  • Added support for reading alarm tokens from cfg.
  • Added support for Web-based Service Oriented Configuration (SOC).
  • Fixed memory leak.
December 2010
3.40
  • Applied fix to resolve the issue: TNT2 metric different for clear alarms.
  • Fixed Desktop Handle leak.
August 2010
3.30
Applied a fix related to process name display on UNIX platforms.
April 2010
3.21
  • Fixed an issue related to process restart alarm.
  • Fixed ci_metric_type for minimum memory usage alarm.
April 2010
3.20
  • Added support for extended NIS database information.
  • Added scheduler support for individual profiles QoS and Alarms.
  • Added support for action command on alarms for CPU profile.
March 2010
3.14
Added support for alarm clear when the monitored process goes down.
March 2010
3.13
Truncated process command line on windows to 1000 bytes.
February 2010
3.12
  • Added a fix to the data collector in the processes probe.
  • Alarm for incorrect instance not sent of the process is in the exclude list.
  • Added the missing MsgCpuUsageMin and MsgCpuUsageRange alarms to the cfx file.
  • Added support for LINUX_22.
  • Overwrite of subsystem ID for all alarms is now possible.
  • Handling of process names greater than 32 characters in Solaris.
  • Script execution on alarm conditions.
  • Fixed the exclude functionality. No alarms or QoS are sent for processes that are excluded.
  • Support for $robot variable in messages.
  • Added 'default' flag for alarm messages.
  • Improved error handling for fetching performance data.
  • Added default clear message. Fixed configuration tool and probe to enable the use of the clear message.
  • Fixed minor GUI bug when using the new grouping of profiles feature. Sometimes, when the user double-clicks a process in the status tab, and that profile is monitored by a profile, the profile would not be displayed unless the group were active and selected in the profiles tab.
October 2009
2.90
  • Added option to group profiles together. Group name can be used as part of alarm messages. Groups can be activated/deactivated without having to activate/deactivate all profiles within a group. Profiles can be moved between groups.
  • Added option to clone a profile (create a profile with same settings as the source).
  • Added option to select many rows in profiles list box.
  • Added option to 'track processes by process ID'. This opens up a new feature, to alert when a process has been restarted. It also opens the possibility to send in individual QoS samples for otherwise similar processes. Processes which you can't typically separate using the standard methods like RegEx or command line arguments.
  • Added option to monitor process instances between a given range.
  • Added option to alert if avg. cpu falls below a given threshold (similar as thread count and memory usage).
  • Added option to invert the test for process owner (to be able to alert when a process is NOT running under a given user). Easier than inverting a RegEx method.
  • Fix: The GUI and the probe were treating the "proc_cmd_line" key differently which caused confusion. Probe accepted the string without the key "scan_proc_cmd_line", while the GUI did not. Now the GUI interprets these keys the same way the probe does.
  • Fix: Improved various GUI input validation fields. For example: text fields which accept only numerical values would now only accept numbers.
  • Improvement: If you are running the processes probe on a Windows system, the probe GUI displays both the working set (VM) and the pagefile memory in two separate columns for processes.
  • Fix: Trailing blank space in command line arguments caused profile matching to fail on Windows platforms.
April 2009
2.73
Rebuild following NimBUS library fixes.
December 2008
2.72
Probe update
November 2008
2.71
  • Fixed configuration tool problems with renamed profiles.
  • Fixed shortcut keys for create and delete profile.
  • Fixed active state for on action input fields.
  • Added support for 64-bit Windows (x64).
  • Added support for LINUX on 64-bit PowerPC (glibc = 2.3).
  • Added support for 64-bit AIX, HP-UX, SOLARIS, and LINUX.
  For version 2.71 and higher of this probe, NimBUS Robot version 3.00 (or higher) is a prerequisite. You are advised to read the document "Upgrading the NimBUS Robot" before installing/upgrading.
September 2008
2.53
Windows: Program failure on > 4k command lines fixed.
October 2007
2.52
  • Windows: Case independence of regular expressions fixed.
  • Windows: Command line process lookup improved for 32-bit processes.
  • Known issues:
  • The probe is unable to get the command line of 64-bit processes on Windows and of all processes on Windows Vista.
  • From version 2.40, it is possible to try to retrieve process information again if the probe believes the data to be corrupt. This limit has been defined to a default value of 1 (which means, try 1 more time and then give up). You can tweak the value in the raw-configure to any number between 0 and 10.
September 2007
2.51
  • UNIX: Added new versions to supported platform list:
  • HP-UX_11: 32-bit (PA-RISC)
  • LINUX: 32-bit (Glibc >=2.3)
  • SOLARIS_10: 32-bit (x86)
  • Windows: Optimized code for fetching CPU usage per process.
May 2007
2.40
  • Fix: Improved logic to try to detect corrupt process information in some rare situations on HP-UX. If the probe detects corrupt data, there is a retry function that tries to retrieve process information again.
  • Fix: The test button in the GUI, for testing a profile against running processes did not respect the 'Excluded processes' list. This has been fixed.
  • Fix: The GUI sometimes did not show if a process was being monitored, even if the probe was monitoring the process. The reason for this was that the GUI was trying to match profiles against the process list from the probe. The process list returned to the GUI (from the probe) was a snapshot of current running processes. The process match against profiles logic has been moved to the probe; so the probe GUI now displays the list of processes that the probe monitors. A process may also be monitored by more than 1 profile, which in previous versions the GUI was unable to handle.
  • GUI: New input field to allow custom log file size.
  • GUI: The help button in the dialog for Profile Monitoring dialog is now pointing to Profile setup in the documentation.
  • Fix: Probe should no longer be case-sensitive on all windows platforms when matching process owner, process name, or command line arguments.
  • GUI: Added a new menu item on the right-click context menu of the listview control that displays profiles. Prompts you to give a new profile name. If you select OK and the name does not already exist, the profile is renamed once you save your configuration file.
  • GUI: The 'Expected Instances' field in the Profile Monitoring dialog has been changed to allow more processes. The limit has been increased from 999 to 9999.
  • Windows: skip command line detection for 'System' process to avoid intermittent memory access violations.
January 2007
2.35
  • Solaris: logging prints errno variable even when no error has occurred depicting a non-existing problem.
  • Unix: Improved handling of child processes (fixes a problem with the signal handler which would cause the probe to hang sometimes)
  • Unix: Improved handling of forked processes
September 2006
2.32
AIX: Fix problem getting the process list on v5.3
February 2006
2.30
Added process recognition by resource.
Message variables are made available to more alarm situations. The current set of variables are: arguments, command, description,  errmsg,  executable,  max_restarts, pid, proc_cmd_line,  process, start_dir, watcher
Instances
  • instances_expect
  • instances_found
  • instances_op
  • process_count
  • process_count_type
  • Processes
  • cpu_average
  • cpu_total
  • expected_cpu_usage
  • expected_size
  • expected_user
  • max_samples
  • op
  • samples
  • size
  • thread_limit
  • threads
  • user
  • time_delta
  • which
  • Window
  • window_name
  • window_class
  • window_text
March 2005
Probe Specific Hardware Requirements
The processes probe should be installed on systems with the following minimum resources:
  • Memory: 2-4GB of RAM. The OOB configuration of the probe requires 256MB of RAM
  • CPU: 3-GHz dual-core processor, 32-bit, or 64-bit
If you install 32-bit OS in a Solaris system with 64-bit architecture, then the probe stops functioning and the status indicator on the GUI turns red.
Probe Specific Software Requirements
The processes probe requires the following software environment:
  • CA Unified Infrastructure Management 8.4 or later
  • Robot 7.70 or later (recommended)
  • (From version 4.64)
    Install robot 7.97 on the hub
  • (to enable FIPS encryption) Bus (Robot) version 7.80
  • Probe Provisioning Manager (ppm) probe version 2.38 or later (required for Admin Console)
  • Java JRE 6 or later (required for Admin Console)
  • (From version 4.32)
    IPv6
The probe also requires the following package to monitor resident memory:
  • (From version 4.64):
    • Install the mps_language_pack 8.6.2 version and then restart
      wasp
      .
    • Install the ci_defn_pack 1.5.2 version and then restart
      nis_server
      probe.
    • Install the wasp language pack 9.0.2-A version and then restart
      wasp
      .
  • Install the
    mps_language_pack
    probe version 8.31 or later. You must restart the
    service_host
    probe after you deploy the
    mps_language_pack
    .
You must install the /usr/ucb package manually with Solaris 11 for running the probe. By default, this package is installed with Solaris 10.
(Version 4.50) Probe Specific Changes After Upgrade
The probe supports AES-128 bit encryption and decryption. To enable this feature in your Infrastructure Manager (IM), set the
NIM_PROBE_CRYPTO_MODE
environment variable to one of the following values in the controller probe:
  • TWOFISH
    (default)
  • AES_128_COMPAT
  • If you do not set any value of the environment variable, then the probe uses
    TWOFISH
    encryption and decryption, by default.
  • Set this value in all the robots where the processes probe is configured.
  • Restart the
    Nimsoft Robot Watcher
    service after you set the environment variable.
  • The probe fails to start if you provide incorrect values for the environment variable. CA does not recommend you to change this value after you set it once.
    However, if your probe still fails to respond, see
    Known Issues and Workarounds
    section for more information.
Upgrade Considerations
  • On upgrading the probe to version 4.20 on Linux and AIX systems, to use the
    binary path
    feature on the existing profiles, you must edit the profiles and specify the binary path.
  • While upgrading the probe version from 3.83 to 3.91, process up and down alarms are not cleared from Alarm Console automatically. You must acknowledge these alarms manually.
  • To view the metrics that are available in the processes probe version 3.9 on the USM portal, perform one of the following actions:
    • Upgrade your NMS version to NMS 7.6 or CA UIM 8.0 or later.
    • Install ci_defn_pack probe version 1.02. You must restart the
      nis_server
      when you deploy the
      ci_defn_pack
      .
(Version 4.70) Template Migration steps for Processes probe:
  • Enhanced processes MCS template 4.64 to Enhanced processes MCS template 4.70.
Known Issues and Workarounds
The known issues of the probe are:
  • (On Version 4.50)
    On your IM on a Windows system, sometimes the probe GUI might not open or might not respond. To fix this issue, create an environment variable on the Windows system from where you are accessing the IM.
    Follow these steps:
    1. Right-click on
      My Computer
      and select
      Properties
      .
    2. Click
      Advanced System Settings
      to open the
      System Properties
      dialog.
    3. Click
      Environment Variables
      , and then click
      New
      to add a new variable.
    4. Specify the following details for the
      New User Variable
      :
      • Variable name
        : NIM_PROBE_CRYPTO_MODE
      • Variable value
        : AES_128_COMPAT
    5. Click
      OK
      and exit from System Properties.
    6. Restart the
      Nimsoft Robot Watcher
      service after you set the environment variable.
  • (AIX and HP-UX platforms)
    Processes that do not execute as soft links are displayed as the command line output that is returned by the API.
The probe has the following known issues around the process name:
  • Two profiles that monitor the same process are assigned the same suppression key (which is based on the process name). The probe generates alarms for both the profiles, but the USM recognizes only one (based on suppression key).
  • The probe alarms are ambiguous if the monitoring profile of two or more processes have similar names. By default, the probe does a regular expression check for processes names. For example, the probe considers NOTEPAD and NOTEPAD++ as similar names and hence generates or clears ambiguous alarms.
    Workaround:
    Select the
    Track Processes by Process Identifier
    check box in the
    General Setup
    section of the probe. When selected, the probe uses the profile name to generate suppression keys, allowing the USM to display the alarms accurately.
  • The 2.40 version of the probe can reattempt to retrieve the process information if there are any indications of corrupt data. By default the probe retries once and this setting can be changed in the
    Raw Configure
    GUI to any value between 0 and 10.
  • The probe generated false alarms in Alarm Console for a profile having multiple processes with different names. For example, you create a profile called nimbus* and select the
    Track Processes by Process Identifier
    check box in the
    General Setup
    section of the probe. When you deactivate one of the nimbus processes, the probe generates false alarms and you may not obtain any information about which process is down. Thus, we recommend creating separate profiles for each process.