Primary Storage

 View Only
  • 1.  IBM Storage subsystem FC ports reset threshold

    Posted Thu June 17, 2021 12:15 PM
    IHAC who made a question.

    Under certain circumstances when in the SAN:
    • misbehaving hosts, for example, become more common as hardware ages,
    • bad host behavior usually caused by defective host bus adapter (HBA) hardware, bugs in the HBA firmware,
    • problems with HBA drivers or Storage ports which produce the same symptoms (high latency) due to defective interface hardware or firmware issues,

    In this situation some arrays deliberately reset their fabric ports if they are not receiving host responses within their specified timeout periods.

    Is there any reference which shows the threshold limits for different subsytems, like DS8K or SV family.

    Thanks  



    IBM Italia S.p.A. Sede Legale: Circonvallazione Idroscalo - 20090 Segrate (MI) Cap. Soc. euro 347.256.998,80 C. F. e Reg. Imprese MI 01442240030 - Partita IVA 10914660153 Societa' con unico azionista Societa' soggetta all'attivita' di direzione e coordinamento di International Business Machines Corporation (Salvo che sia diversamente indicato sopra / Unless stated otherwise above)


  • 2.  RE: IBM Storage subsystem FC ports reset threshold

    Posted Fri June 18, 2021 05:59 AM

    Ciao Angelo,

    from a SpecV point of view I doubt we take any recovery action concerning unresponsive hosts.
    I rather believe, by design we consider the host as offline and we (a SpecV system) need nothing do about it.
    Not sure though if it would make a difference whether the host went offline gracefully and we received an SCN from the SAN switch or it went away silently - or simply does not respond because it hangs or similar.
    Last not least, AFAIK we as SCSI target would not trigger recovery actions as port reset / bus reset etc.;
    resetting our own fabric ports would impact any other ongoing I/O.

    Just my thoughts on that...




    ------------------------------
    Christian Schroeder
    IBM SpecV Storage Support with Passion
    ------------------------------



  • 3.  RE: IBM Storage subsystem FC ports reset threshold

    IBM Champion
    Posted Fri June 18, 2021 06:24 AM
    Edited by Nezih Boyacioglu Fri June 18, 2021 06:25 AM
    Brocade FOS9 has a new feature called FPIN (Fabric Performance Impact Notification) to notify host during congestion, over subscription, credit stall or mpio issues. But the most of the operating systems is not ready to understand the fpin notification and take an action.  I don't think the storage admins wants to take automatic action to disable port or something like that before investigation. 

    About FPIN:
    https://docs.broadcom.com/doc/FOS-90-Fabric-Notifications-OT

    ------------------------------
    NEZIH BOYACIOGLU
    ------------------------------



  • 4.  RE: IBM Storage subsystem FC ports reset threshold

    Posted Mon June 21, 2021 12:18 PM
    For writes only, if a host has not responded to a transfer request within 12 seconds the command is internally aborted by the node. If this happens three times on the same login, the login is closed by the node causing it to reset.

    ------------------------------
    Chris Bulmer
    ------------------------------