perfSONAR User Q&A and Other Discussion

[Mark Feit <>]

(Very Short Version:  Start using the label parameter in your archive specifications to make diagnostics easier.)

 

 

Several years ago, we added a label parameter to pScheduler’s archive specification to allow for better diagnostics when examining the results of individual measurements:

 

$ pscheduler result --full https://host/pscheduler/...

…

Archivings:

 

  To Fail-o-Matic 5000 using failer, Unfinished

    2024-07-22T16:49:38Z      Induced random failure, not retrying

 

  To Central Archive using http, Finished

    2024-07-22T16:49:38Z      Succeeded

 

  To http, Finished                        These two are different and hard to tell apart.

    2024-07-22T16:49:38Z      Succeeded

 

  To http, Finished

    2024-07-22T16:49:38Z      Succeeded

 

 

In 5.2.0 (to be released later this year), pScheduler will include a new command  to produce a summary of the success and failure rates of archivings over a recent period of time:

 

$ pscheduler archiving-summary PT1H

 

Succeeded        Failed            Label

---------------  ----------------  -------------------------

      10 (100%)          0 (  0%)  ABC University archive

       0 (  0%)          5 (100%)  University of DEF archive

      12 ( 57%)          9 ( 43%)  GHI Institute archive

      26 (100%)          0 (  0%)  syslog

      23 (100%)          0 (  0%)  syslog #2

---------------  ----------------  -------------------------

      71 ( 84%)         14 ( 16%)  Total (85)

 

The Label  column, which is what makes each row unique,  is based on the label specified in the archive specification or, if none is specified, the name of the archiver plugin used (e.g., syslog in the example above).  When the data in the archive specification is different but the label or plugin is the same, the report will append a number to differentiate them  (e.g., syslog and syslog #2 in the example above). 

 

Note that, because this information is pulled from the pScheduler API, sensitive information in the archive specification is not considered when determining uniqueness.  Without labels, these two archive specifications would be treated as being the same and combined in the statistics:

 

{

    "archiver": "http",

    "data": {

        "_url": https://archive.this.edu/logstash,    (Sensitive)

        "op": "post"

    }

}

 

{

    "archiver": "http",

    "data": {

        "_url": https://archive.that.edu/logstash,    (Sensitive)

        "op": "post"

    }

}

 

We’d like to recommend that everyone adopt the practice of adding unique, descriptive labels to their archiving specifications to make them easier to tell apart:

 

{

    "schema": 3,

    "label": "This University Central Archive",

    "archiver": "http",

    "data": {

        "_url": "https://archive.this.edu/logstash",

        "op": "post"

    }

}

 

{

    "schema": 3,

    "label": "That University Central Archive",

    "archiver": "http",

    "data": {

        "_url": "https://archive.that.edu/logstash",

        "op": "post"

    }

}

 

Specifying a schema version of 3 or higher required for pScheduler to understand labels.

 

 

As always, if you have questions, don’t hesitate to reply to the list or drop us a line.

 

--Mark