Anubis not working in SM application cluster ???

Hello,

even after enabling new anubis option in sm.ini file when a background process dies anubis doesnt start it. We have 4 application servers  in application group cluster.  Background processes are started from sm.cfg on servers (on Server1 there is agent, lister, scheduledAction etc.; on Server2 there is linker, sla and so on). It seems that new anubis is not able to start background processes in this architecture.... please any tip or trick :) We have SM 9.41 RTE in place. Kerob

  • Helo Kerob,

       After enabling the enableAnubisMonitor and anubisPollInterval on the sm.ini, did you disabled the original RAD applications?

    Important: The ThreadControllerId-background thread, which can be started no matter whether enableAnubisMonitor is enabled or not, has been modified to additionally include theanubis monitor functionality and thus eliminated the need for the anubis process. You are recommended to use the new anubis parameters instead of the old anubis process. If an anubisagent record and anubis schedule record, which were used to restart stopped background processes automatically, already exist in your system, before enabling enableAnubisMonitoryou need to disable the original anubis RAD applications. To do this, follow these steps:

    1. Delete the existing anubis agent record.
      1. Type info on the Service Manager command line, and then click Search.
      2. Delete the anubis agent record.
    2. Delete the existing anubis schedule record.
      1. Type sch on the Service Manager command line, and then click Search.
      2. Delete the anubis schedule records 

    Regards,

  • Hello,

    sure. The "old" anubis way is disabled. Thx.

  • The anubis process can only restart background processes started by record "startup" in "info" dbdict.

    The command "sm system.start" in sm.cfg is actually executing this startup record.

    As this command starts a RAD application inside a new process, it spawns the background schedulers as threads in this process, This means all background schedulers started this way, run as threads in the same process - and on the same machine.

    You can alternatively start each background scheduler in separate process by starting it from command line or sm.cfg file like this:

    ./sm -bg scheduler linker -log:../logs/linker.log

    This has some advantages like each process having all of the process virtual address space available without sharing, or that you can specify a separate log file, or specify trace parameters for it on command line. However, anubis is not designed to monitor background schedulers started this way.

     

  • Hello,

    uff scaring ! We must start background schedulers in sm.cfg because of balancing load across SM servers in SM application cluster. On each server there is in its sm.cfg section stating which background processes to be started on this particular server...

    It means, that anubis cannot monitor and restart background processes that were started from sm.cfg ?

    And is there any way how to say "please start this background process on this server and another background process on that server"  in record "startup" in "info" dbdict ? If not then all background processes are started on one server (of 4 servers) and this is problem from performance and fail-over views. 

    Thank you in advance . Kerob

     

  • The last question... on that... :) I hope.

    Is anubis (either the old or the new rte one) able to monitor/restart background processes which run on different servers within one SM application cluster ? And if so, please how ?

    Thank you

    Kerob

  • The following two param helps eliminate the need for anubis process. Please check online help ( copied the following from online help)

    Startup parameter: enableAnubisMonitor:

    Startup parameters change the behavior of the HP Service Manager server. You can always set a startup parameter from the server's operating system command prompt.

    Parameter

    enableAnubisMonitor

    Description

     

    This parameter starts the anubis monitor to check the status of background scheduler threads. By default, this parameter is disabled.

    This parameter should be used in conjunction with the anubisPollInterval parameter, which specifies the time interval (in seconds) at which the anubis monitor checks to see if any registered scheduler is alive (default: 15 seconds).

    Important: The ThreadControllerId-background thread, which can be started no matter whether enableAnubisMonitor is enabled or not, has been modified to additionally include the anubis monitor functionality and thus eliminated the need for the anubis process. You are recommended to use the new anubis parameters instead of the old anubis process. If an anubis agent record and anubis schedule record, which were used to restart stopped background processes automatically, already exist in your system, before enabling enableAnubisMonitor you need to disable the original anubis RAD applications. To do this, follow these steps:

    1. Delete the existing anubis agent record.
      1. Type info on the Service Manager command line, and then click Search.
      2. Delete the anubis agent record.
    2. Delete the existing anubis schedule record.
      1. Type sch on the Service Manager command line, and then click Search.
      2. Delete the anubis schedule record.

     

    Valid if set from

    Server's operating system command prompt

    Initialization file (sm.ini)

    Requires restart of the Service Manager server?

    Yes

    Default value

    0 (Disabled)

    Possible values

    0: Disabled (when this parameter is not present in sm.ini or the command prompt)

    1: Enabled

    Example usage

    Command line: sm -httpPort:13080 -enableAnubisMonitor:1

    Initialization file: enableAnubisMonitor:1

    Startup parameter: anubisPollInterval

    Startup parameters change the behavior of the HP Service Manager server. You can always set a startup parameter from the server's operating system command prompt.

    Parameter

    anubisPollInterval

    Description

     

    This parameter specifies the time interval (in seconds) at which the anubis monitor checks to see if any registered scheduler is alive.

    This parameter should be used in conjunction with the enableAnubisMonitor parameter.

    Important: The ThreadControllerId-background thread, which can be started no matter whether enableAnubisMonitor is enabled or not, has been modified to additionally include the anubis monitor functionality and thus eliminated the need for the anubis process. You are recommended to use the new anubis parameters instead of the old anubis process. If an anubis agent record and anubis schedule record, which were used to restart stopped background processes automatically, already exist in your system, before enabling enableAnubisMonitor you need to disable the original anubis RAD applications as follows:

    1. Delete the existing anubis agent record.
      1. Type info on the Service Manager command line, and click Search.
      2. Delete the anubis agent record.
    2. Delete the existing anubis schedule record.
      1. Type sch on the Service Manager command line, and click Search.
      2. Delete the anubis schedule record.

     

    Valid if set from

    Server's operating system command prompt

    Initialization file (sm.ini)

    Requires restart of the Service Manager server?

    No

    Default value

    15 (seconds)

    Possible values

    Number of seconds

    Example usage

    Command line: sm -httpPort:13080 -anubisPollInterval:15

    Initialization file: anubisPollInterval:15

  • The following two param helps eliminate the need for anubis process. Please check online help ( copied the following from online help)

    Startup parameter: enableAnubisMonitor:

    Startup parameters change the behavior of the HP Service Manager server. You can always set a startup parameter from the server's operating system command prompt.

    Parameter

    enableAnubisMonitor

    Description

     

    This parameter starts the anubis monitor to check the status of background scheduler threads. By default, this parameter is disabled.

    This parameter should be used in conjunction with the anubisPollInterval parameter, which specifies the time interval (in seconds) at which the anubis monitor checks to see if any registered scheduler is alive (default: 15 seconds).

    Important: The ThreadControllerId-background thread, which can be started no matter whether enableAnubisMonitor is enabled or not, has been modified to additionally include the anubis monitor functionality and thus eliminated the need for the anubis process. You are recommended to use the new anubis parameters instead of the old anubis process. If an anubis agent record and anubis schedule record, which were used to restart stopped background processes automatically, already exist in your system, before enabling enableAnubisMonitor you need to disable the original anubis RAD applications. To do this, follow these steps:

    1. Delete the existing anubis agent record.
      1. Type info on the Service Manager command line, and then click Search.
      2. Delete the anubis agent record.
    2. Delete the existing anubis schedule record.
      1. Type sch on the Service Manager command line, and then click Search.
      2. Delete the anubis schedule record.

     

    Valid if set from

    Server's operating system command prompt

    Initialization file (sm.ini)

    Requires restart of the Service Manager server?

    Yes

    Default value

    0 (Disabled)

    Possible values

    0: Disabled (when this parameter is not present in sm.ini or the command prompt)

    1: Enabled

    Example usage

    Command line: sm -httpPort:13080 -enableAnubisMonitor:1

    Initialization file: enableAnubisMonitor:1

    Startup parameter: anubisPollInterval

    Startup parameters change the behavior of the HP Service Manager server. You can always set a startup parameter from the server's operating system command prompt.

    Parameter

    anubisPollInterval

    Description

     

    This parameter specifies the time interval (in seconds) at which the anubis monitor checks to see if any registered scheduler is alive.

    This parameter should be used in conjunction with the enableAnubisMonitor parameter.

    Important: The ThreadControllerId-background thread, which can be started no matter whether enableAnubisMonitor is enabled or not, has been modified to additionally include the anubis monitor functionality and thus eliminated the need for the anubis process. You are recommended to use the new anubis parameters instead of the old anubis process. If an anubis agent record and anubis schedule record, which were used to restart stopped background processes automatically, already exist in your system, before enabling enableAnubisMonitor you need to disable the original anubis RAD applications as follows:

    1. Delete the existing anubis agent record.
      1. Type info on the Service Manager command line, and click Search.
      2. Delete the anubis agent record.
    2. Delete the existing anubis schedule record.
      1. Type sch on the Service Manager command line, and click Search.
      2. Delete the anubis schedule record.

     

    Valid if set from

    Server's operating system command prompt

    Initialization file (sm.ini)

    Requires restart of the Service Manager server?

    No

    Default value

    15 (seconds)

    Possible values

    Number of seconds

    Example usage

    Command line: sm -httpPort:13080 -anubisPollInterval:15

    Initialization file: anubisPollInterval:15

  • O my God ! All done many times. No sense to copy help topics here.  It seems, that all thi stuff works only on simple vertical architecture and not on complex (enterprise scale) horizontal architecture where multiple application servers work in application server group... (thus only particular background processes are running on particular nodes (servers) on that group).

    Lines from sm.log

    4796(  4800) 01/29/2016 19:27:09 JRTE I Anubis monitor is enabled. Please make sure that the anubis RAD apps should be disabled.
      4796(  4800) 01/29/2016 19:27:09 JRTE I Anubis polling interval is 20 seconds.
      4796(  4800) 01/29/2016 19:27:09  RTE I Host network address: 10.7.17.100

  • hello, 

    Didn't mean to overwhelm you with online information. The HZ scaled environment is being tested. We will post a reply shortly.

    thanks,

    Rama

  • From the log file I can tell that you have the new anubis configured correctly.  I suspect that in your sm.cfg file you have statements like:

    sm scheduler linker -bg

    sm schduler problem -bg

    The new anubis was written to only monitor two applications,  "system.start" and "scheduler.start".   So currently in order for it to work with individual schedulers in the sm.cfg you would need to modify the lines to look like

    sm scheduler.start linker.startup -bg

    sm scheduler.start problem -bg

    where "linker.startup" and "problem" are the names of the info,startup records in the info table.

    I agree this is less than an optimal design and i will open a case with the R&D team to see if it is possible to change anubis to also montior application "scheduler" so that customers are not required to change the sm.cfg file.

    If I have misunderstood your sm.cfg file please let me know.

    Last the new anubus behaves the same in a veritical or horizontal deployment.

    Steve Hirschfeld

    I