Uploaded image for project: 'Help-Desk'
  1. Help-Desk
  2. HELP-5820

FIWARE.Request.Tech.Data.CEP.Proton load issue

    Details

    • Type: extRequest
    • Status: Closed
    • Priority: Major
    • Resolution: Done
    • Fix Version/s: 2021
    • Component/s: FIWARE-TECH-HELP
    • Labels:
      None

      Description

      Hi,

      I am forwarding this message (as suggested) to report a Proton issue.

      Thanks,

      MarcoF

      ---------- Forwarded message ----------
      From: Fabio Lanari (Kiunsys) <fabio.lanari@kiunsys.com>
      Date: 2 February 2016 at 08:51
      Subject: Fwd: RE: Opportunities for dissemination in FIWARE events
      To: Marco Fagiani <m.fagiani85@gmail.com>
      Cc: Fabio Formosa <f.formosa@liberologico.com>

      La risposta di Telefonica.

      -------- Messaggio Inoltrato --------
      Oggetto: RE: Opportunities for dissemination in FIWARE events
      Data: Mon, 1 Feb 2016 22:54:32 +0000
      Mittente: SERGIO GARCIA GOMEZ <sergio.garciagomez@telefonica.com>
      <sergio.garciagomez@telefonica.com>
      A: Fabio Lanari (Kiunsys) <fabio.lanari@kiunsys.com>
      <fabio.lanari@kiunsys.com>, JOSE MANUEL CANTERA FONSECA
      <josemanuel.canterafonseca@telefonica.com>
      <josemanuel.canterafonseca@telefonica.com>, Paola Ponticelli
      <paola.ponticelli@kiunsys.com> <paola.ponticelli@kiunsys.com>
      CC: SANTIAGO MARTINEZ GARCIA <santiago.martinezgarcia@telefonica.com>
      <santiago.martinezgarcia@telefonica.com>

      Dear Fabio,

      The best way to handle this is to forward this message to
      fiware-tech-help@lists.fiware.org

      It’s better if you do it because you will get the email notifications on
      changes and answers. We will notify the Proton team anyway.

      Best Regards,

      Sergio.

      De: Fabio Lanari (Kiunsys) [mailto:fabio.lanari@kiunsys.com
      <fabio.lanari@kiunsys.com>]
      Enviado el: 01 February 2016 10:24
      Para: JOSE MANUEL CANTERA FONSECA
      <josemanuel.canterafonseca@telefonica.com>
      <josemanuel.canterafonseca@telefonica.com>; Paola Ponticelli
      <paola.ponticelli@kiunsys.com> <paola.ponticelli@kiunsys.com>
      CC: SERGIO GARCIA GOMEZ <sergio.garciagomez@telefonica.com>
      <sergio.garciagomez@telefonica.com>
      Asunto: Re: Opportunities for dissemination in FIWARE events

      Hi Sergio,
      we have used these few days for performing one more test (VM scaling up).

      The main problem, as I said, is that Proton is consuming too much HW
      resources while managing only a bunch of alerting rules, which produces or
      the app crash or requires the VM restart.

      We have tested the entire system in 2 different IaaS installations.

      a) GEs used: IDAS IoT Agent 1.3.1. + ORION 0.26.1 + PROTON CEP 3.3.1
      b) Installation type: local installation on a IaaS provider
      c) Installation method: Docker + Docker Compose

      Installations specs:

      • 1st installation: 1 VM (14 GB RAM, 4 core) containing the 3
        GEs
        <-- in this case we haven' seen any problems. It should be said that this
        was an initial installation where we implemented only 2 or 3 Proton rules
      • 2nd installation - small size (due to IaaS provider migration): 1 VM (4
        GB RAM, 2 core) containig the 3 GEs <-- in this case Proton eats up to 1.2
        GB of RAM and constantly uses 98% of 2 cores (with 8 Proton rules)
      • 2nd installation - mid-size (scaling up of the previous VM): 1 VM (8 GB
        RAM, 4 core) containing the 3 GEs <-- Proton still eats up to 1.2 GB of
        RAM and constantly uses 3 cores at 100% (with 8 Proton rules)

      We don't know where the problem is.
      Our dev team thought there was too much messages between Orion and Proton,
      that Proton is unable to manage, so they are looking for a workaround, a
      method to reduce them. Here is the link to our bug/request:

      http://stackoverflow.com/questions/34491828/fiware-orion-context-broker-metadata-updates-trigger-notifications

      Thank you for your help

      Fabio L.

      -------- Messaggio originale --------
      Oggetto: Re: Opportunities for dissemination in FIWARE events
      Mittente: JOSE MANUEL CANTERA FONSECA
      <josemanuel.canterafonseca@telefonica.com>
      <josemanuel.canterafonseca@telefonica.com>
      <josemanuel.canterafonseca@telefonica.com>
      A: Paola Ponticelli <paola.ponticelli@kiunsys.com>
      <paola.ponticelli@kiunsys.com>
      Data: 29/1/2016 10:13:56

      Fabio, Paola,

      Please don’t forget to send us the ticket and some description about the
      problems found on your integration.

      Thanks, best

      Since January 1st, old domains won't be supported and messages sent to any domain different to @lists.fiware.org will be lost.
      Please, send your messages using the new domain (Fiware-tech-help@lists.fiware.org) instead of the old one.
      _______________________________________________
      Fiware-tech-help mailing list
      Fiware-tech-help@lists.fiware.org
      https://lists.fiware.org/listinfo/fiware-tech-help
      [Created via e-mail received from: Marco Fagiani <m.fagiani85@gmail.com>]

      1. INES.json
        21 kB
        FW External User

        Activity

        Hide
        TALI Fabiana Fournier added a comment -

        The CEP load is determined by several factors:
        1. The incoming events rate
        2. The output event rate
        3. The number of rules
        4. The number of open processing contexts that are open in parallel during a run
        5. The complexity of the rules evaluation

        I suspect that the issue here relates to #4, but in order to analyse it I need more data:
        1. The CEP definition file (can be exported from the Authoring Tool)
        2. The input events that were processed by the CEP and their approximate volume per second

        Thanks

        Show
        TALI Fabiana Fournier added a comment - The CEP load is determined by several factors: 1. The incoming events rate 2. The output event rate 3. The number of rules 4. The number of open processing contexts that are open in parallel during a run 5. The complexity of the rules evaluation I suspect that the issue here relates to #4, but in order to analyse it I need more data: 1. The CEP definition file (can be exported from the Authoring Tool) 2. The input events that were processed by the CEP and their approximate volume per second Thanks
        Hide
        TALI Fabiana Fournier added a comment -

        The CEP load is determined by several factors:
        1. The incoming events rate
        2. The output event rate
        3. The number of rules
        4. The number of open processing contexts that are open in parallel during a run
        5. The complexity of the rules evaluation
        I suspect that the issue here relates to #4, but in order to analyse it I need more data:
        1. The CEP definition file (can be exported from the Authoring Tool)
        2. The input events that were processed by the CEP and their approximate volume per second
        Thanks,
        Tali

        Show
        TALI Fabiana Fournier added a comment - The CEP load is determined by several factors: 1. The incoming events rate 2. The output event rate 3. The number of rules 4. The number of open processing contexts that are open in parallel during a run 5. The complexity of the rules evaluation I suspect that the issue here relates to #4, but in order to analyse it I need more data: 1. The CEP definition file (can be exported from the Authoring Tool) 2. The input events that were processed by the CEP and their approximate volume per second Thanks, Tali
        Hide
        fw.ext.user FW External User added a comment -

        Hi,

        Two events (PSSContextUpdate and DCUContextUpdate) are defined, with the
        same 11 attributes, in addition to the entityId and entityType.
        Each minute, about 3 DCU events, and 73 PSS events are received.
        Please find attached the adopted definition file.

        Thanks,

        MarcoF

        Show
        fw.ext.user FW External User added a comment - Hi, Two events (PSSContextUpdate and DCUContextUpdate) are defined, with the same 11 attributes, in addition to the entityId and entityType. Each minute, about 3 DCU events, and 73 PSS events are received. Please find attached the adopted definition file. Thanks, MarcoF
        Hide
        TALI Fabiana Fournier added a comment -

        Thank you Marco for the details.

        An initial investigation of your application definition didn't reveal why the CEP uses a lot of resources (all contexts but one uses "ignore" type initiator, one uses "add" but it is reported to have low frequency of about 3 events per minute, and it has relative 30 minutes terminator).

        I open a bug ticket to have our developer investigate it:
        FIWARE.Bug.Data.CEP.HeavyLoad.Analyze
        https://jira.fiware.org/browse/DATA-1324

        Thanks,
        Tali

        Show
        TALI Fabiana Fournier added a comment - Thank you Marco for the details. An initial investigation of your application definition didn't reveal why the CEP uses a lot of resources (all contexts but one uses "ignore" type initiator, one uses "add" but it is reported to have low frequency of about 3 events per minute, and it has relative 30 minutes terminator). I open a bug ticket to have our developer investigate it: FIWARE.Bug.Data.CEP.HeavyLoad.Analyze https://jira.fiware.org/browse/DATA-1324 Thanks, Tali

          People

          • Assignee:
            TALI Fabiana Fournier
            Reporter:
            fw.ext.user FW External User
          • Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved: