DMP VXVM 4.0 and Array HP XP1024 - Veritas Volume Manager

This is a discussion on DMP VXVM 4.0 and Array HP XP1024 - Veritas Volume Manager ; Hi Gurus: I have a problem with vxdmp en VXVM 4.0 . I will try my configuracion: Domain SunFire 25K with qlogic 2300 HBA card and connect to Array HP XP1024 . I installed the HPXP1 to support this array ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: DMP VXVM 4.0 and Array HP XP1024

  1. DMP VXVM 4.0 and Array HP XP1024


    Hi Gurus:

    I have a problem with vxdmp en VXVM 4.0 .

    I will try my configuracion: Domain SunFire 25K with qlogic 2300 HBA card
    and connect to Array HP XP1024 . I installed the HPXP1 to support this array
    but when I disconnect a fibre cable the vxdmp don't disable the controller,
    I continue see the controller enable and the domain is hung:
    vxdmpadm listctlr all
    CTLR-NAME ENCLR-TYPE STATE ENCLR-NAME
    ================================================== ===
    c1 Disk ENABLED Disk
    c0 Disk ENABLED Disk
    c4 XP1024 ENABLED XP10240
    c5 XP1024 ENABLED XP10240

    Do you have any ideas? . I have modified the /kernel/drv/vxdmp.conf with
    this line:
    dmp_jbod="HITACHI",
    "HP";
    Also I have modified the vid with this command:
    vxddladm addjbod vid=HP

    Best Regards, Juanma.


  2. Re: DMP VXVM 4.0 and Array HP XP1024

    Did you not see any messages in the /var/adm/messages file ?


    Any scsi errors followed by dmp messages about nodes being disabled ?



    So, here is what DMP does.

    It will send IO down both paths of the XP1024 (If the IO is big enough -
    DMP will load balance). If IO fails (dmp sends IO to the scsi or SD
    level, which sends it to the HBA, which sends it to the XP1024), you
    will first see the scsi level reportnig it (through messages in the
    syslog) and then DMP reporting it, and taking action.

    Once the path is disabled, you will be able to see this.


    Now, your next question will be "how long will this take ?"

    By default the HBA/scsi driver will take about 6 minutes to report a
    failure back to DMP. So wait for at least 6 minutes while doing IO to
    the disks so that a failure can happen.


    Is this what you did ?

    Juan Manuel Sanchez wrote:
    > Hi Gurus:
    >
    > I have a problem with vxdmp en VXVM 4.0 .
    >
    > I will try my configuracion: Domain SunFire 25K with qlogic 2300 HBA card
    > and connect to Array HP XP1024 . I installed the HPXP1 to support this array
    > but when I disconnect a fibre cable the vxdmp don't disable the controller,
    > I continue see the controller enable and the domain is hung:
    > vxdmpadm listctlr all
    > CTLR-NAME ENCLR-TYPE STATE ENCLR-NAME
    > ================================================== ===
    > c1 Disk ENABLED Disk
    > c0 Disk ENABLED Disk
    > c4 XP1024 ENABLED XP10240
    > c5 XP1024 ENABLED XP10240
    >
    > Do you have any ideas? . I have modified the /kernel/drv/vxdmp.conf with
    > this line:
    > dmp_jbod="HITACHI",
    > "HP";
    > Also I have modified the vid with this command:
    > vxddladm addjbod vid=HP
    >
    > Best Regards, Juanma.
    >


  3. Re: DMP VXVM 4.0 and Array HP XP1024


    Hi Guru:

    In the file messages we can see errors with that:

    Jun 6 12:52:44 eup1iq vxdmp: [ID 736771 kern.notice] NOTICE: VxVM
    vxdmp V-5-0-148 enabled path 118/0x8 belonging to the dmpnode 275
    /0x38
    Jun 6 12:52:44 eup1iq vxdmp: [ID 736771 kern.notice] NOTICE: VxVM
    vxdmp V-5-0-148 enabled path 118/0x10 belonging to the dmpnode 27
    5/0x30
    Jun 6 12:56:31 eup1iq qlc: [ID 686697 kern.info] NOTICE: Qlogic qlc
    (0): Loop OFFLINE
    Jun 6 12:56:33 eup1iq qlc: [ID 686697 kern.info] NOTICE: Qlogic qlc
    (0): Link ONLINE
    Jun 6 12:56:33 eup1iq qlc: [ID 686697 kern.info] NOTICE: Qlogic qlc
    (0): Loop OFFLINE
    Jun 6 12:57:42 eup1iq scsi: [ID 243001 kern.warning]
    WARNING: /pci@15c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8003ec1e12,1
    (ssd4):
    Jun 6 12:57:42 eup1iq SCSI transport failed: reason 'timeout':
    retrying command
    Jun 6 12:58:03 eup1iq fctl: [ID 517869 kern.warning] WARNING: 91=>fp
    (0)::OFFLINE timeout
    Jun 6 12:58:22 eup1iq scsi: [ID 243001
    kern.info] /pci@15c,600000/SUNW,qlc@1/fp@0,0 (fcp0):
    Jun 6 12:58:22 eup1iq offlining lun=2 (trace=0), target=31b00
    (trace=2800004)
    Jun 6 12:58:22 eup1iq scsi: [ID 243001
    kern.info] /pci@15c,600000/SUNW,qlc@1/fp@0,0 (fcp0):
    Jun 6 12:58:22 eup1iq offlining lun=1 (trace=0), target=31b00
    (trace=2800004)
    Jun 6 12:58:22 eup1iq scsi: [ID 243001
    kern.info] /pci@15c,600000/SUNW,qlc@1/fp@0,0 (fcp0):
    Jun 6 12:58:22 eup1iq offlining lun=0 (trace=0), target=31b00
    (trace=2800004)
    Jun 6 12:58:22 eup1iq genunix: [ID 408114
    kern.info] /pci@15c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8003ec1e12,2
    (ssd3) offline
    Jun 6 12:58:22 eup1iq genunix: [ID 408114
    kern.info] /pci@15c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8003ec1e12,0
    (ssd5) offline
    Jun 6 13:05:56 eup1iq /usr/lib/snmp/snmpdx: [ID 702911 daemon.error]
    Agent snmpd appeared dead but responded to ping
    Jun 6 13:06:04 eup1iq last message repeated 1 time
    Jun 6 13:17:48 eup1iq scsi: [ID 107833 kern.warning]
    WARNING: /pci@17c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8003ec1e02,2
    (ssd0):
    Jun 6 13:17:48 eup1iq corrupt label - wrong magic number
    Jun 6 13:17:48 eup1iq scsi: [ID 365881 kern.info] Vendor 'HP',
    product 'OPEN-E', 28452960 512 byte blocks
    Jun 6 13:17:48 eup1iq scsi: [ID 107833 kern.warning]
    WARNING: /pci@17c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8003ec1e02,2
    (ssd0):
    Jun 6 13:17:48 eup1iq corrupt label - wrong magic number
    Jun 6 13:17:48 eup1iq scsi: [ID 365881 kern.info] Vendor 'HP',
    product 'OPEN-E', 28452960 512 byte blocks

    You can see that it appers scsi errors but we don't see any messages
    disabled the path.

    I disabled the MPXIO in the driver qlc.conf with this line:
    name="qlc" parent="/pci@15c,600000/SUNW,qlc@1" unit-address="1" mpxio-
    disable="yes";
    name="qlc" parent="/pci@17c,600000/SUNW,qlc@1" unit-address="1" mpxio-
    disable="yes";

    But I think that it's necesary any option in this driver to send the
    error to vxdmp.
    I found a HP document where it comment that it's necesary enable a
    System Option Mode 244 to ON in the XP1024 with Sun HBA. żDo you know
    that make this option?. The XP1024 is working with other Solaris but
    It work with VXVM 3.5 and JNI HBA card's.

    Best Regards, Juanma.


    Me wrote:
    >Did you not see any messages in the /var/adm/messages file ?
    >
    >
    >Any scsi errors followed by dmp messages about nodes being disabled ?
    >
    >
    >
    >So, here is what DMP does.
    >
    >It will send IO down both paths of the XP1024 (If the IO is big enough -


    >DMP will load balance). If IO fails (dmp sends IO to the scsi or SD
    >level, which sends it to the HBA, which sends it to the XP1024), you
    >will first see the scsi level reportnig it (through messages in the
    >syslog) and then DMP reporting it, and taking action.
    >
    >Once the path is disabled, you will be able to see this.
    >
    >
    >Now, your next question will be "how long will this take ?"
    >
    >By default the HBA/scsi driver will take about 6 minutes to report a
    >failure back to DMP. So wait for at least 6 minutes while doing IO to
    >the disks so that a failure can happen.
    >
    >
    >Is this what you did ?
    >
    >Juan Manuel Sanchez wrote:
    >> Hi Gurus:
    >>
    >> I have a problem with vxdmp en VXVM 4.0 .
    >>
    >> I will try my configuracion: Domain SunFire 25K with qlogic 2300 HBA card
    >> and connect to Array HP XP1024 . I installed the HPXP1 to support this

    array
    >> but when I disconnect a fibre cable the vxdmp don't disable the controller,
    >> I continue see the controller enable and the domain is hung:
    >> vxdmpadm listctlr all
    >> CTLR-NAME ENCLR-TYPE STATE ENCLR-NAME
    >> ================================================== ===
    >> c1 Disk ENABLED Disk
    >> c0 Disk ENABLED Disk
    >> c4 XP1024 ENABLED XP10240
    >> c5 XP1024 ENABLED XP10240
    >>
    >> Do you have any ideas? . I have modified the /kernel/drv/vxdmp.conf with
    >> this line:
    >> dmp_jbod="HITACHI",
    >> "HP";
    >> Also I have modified the vid with this command:
    >> vxddladm addjbod vid=HP
    >>
    >> Best Regards, Juanma.
    >>



  4. Re: DMP VXVM 4.0 and Array HP XP1024


    Hi:

    Yestarday , it seems , the problem is solved. It was necesary uninstall the
    last version of SAN Foundait kit SAN.4.4.5 and install the before release:
    SAN.4.4.4 and with this version when a process access to hardisk in disconnect
    controller the vxdmp disable the failing path. Only disable the path hardisk
    because the controller is enable. The controller is disabled only if the
    all hardisk of this controller is failing. The process later about tree
    minutes since the OFFLINE messages apper in the messages file. I will trie
    to modified this time.

    Best regards, Juanma.
    Me wrote:
    >Did you not see any messages in the /var/adm/messages file ?
    >
    >
    >Any scsi errors followed by dmp messages about nodes being disabled ?
    >
    >
    >
    >So, here is what DMP does.
    >
    >It will send IO down both paths of the XP1024 (If the IO is big enough -


    >DMP will load balance). If IO fails (dmp sends IO to the scsi or SD
    >level, which sends it to the HBA, which sends it to the XP1024), you
    >will first see the scsi level reportnig it (through messages in the
    >syslog) and then DMP reporting it, and taking action.
    >
    >Once the path is disabled, you will be able to see this.
    >
    >
    >Now, your next question will be "how long will this take ?"
    >
    >By default the HBA/scsi driver will take about 6 minutes to report a
    >failure back to DMP. So wait for at least 6 minutes while doing IO to
    >the disks so that a failure can happen.
    >
    >
    >Is this what you did ?
    >
    >Juan Manuel Sanchez wrote:
    >> Hi Gurus:
    >>
    >> I have a problem with vxdmp en VXVM 4.0 .
    >>
    >> I will try my configuracion: Domain SunFire 25K with qlogic 2300 HBA card
    >> and connect to Array HP XP1024 . I installed the HPXP1 to support this

    array
    >> but when I disconnect a fibre cable the vxdmp don't disable the controller,
    >> I continue see the controller enable and the domain is hung:
    >> vxdmpadm listctlr all
    >> CTLR-NAME ENCLR-TYPE STATE ENCLR-NAME
    >> ================================================== ===
    >> c1 Disk ENABLED Disk
    >> c0 Disk ENABLED Disk
    >> c4 XP1024 ENABLED XP10240
    >> c5 XP1024 ENABLED XP10240
    >>
    >> Do you have any ideas? . I have modified the /kernel/drv/vxdmp.conf with
    >> this line:
    >> dmp_jbod="HITACHI",
    >> "HP";
    >> Also I have modified the vid with this command:
    >> vxddladm addjbod vid=HP
    >>
    >> Best Regards, Juanma.
    >>



  5. VCS_AGFW[11185]: [ID 702911 daemon.notice] VCS ERROR.


    This is regarding Kernel notice Errors
    VCS_AGFW[11185]: [ID 702911 daemon.notice] VCS ERROR.
    Had[8970]: [ID 702911 daemon.notice] VCS ERROR V-16-.

    vxdmp: [ID 736771 kern.notice] NOTICE: VxVM vxdmp V-0

    eb 15 02:02:59 tcppdbs005 offline
    Feb 15 02:02:59 tcppdbs005 vxdmp: [ID 917986 kern.notice] NOTICE: VxVM vxdmp
    V-0
    Feb 15 02:02:59 tcppdbs005 vxdmp: [ID 824220 kern.notice] NOTICE: VxVM vxdmp
    V-0
    Feb 15 02:04:32 tcppdbs005 VCS_AGFW[11185]: [ID 702911 daemon.notice] VCS
    ERROR.
    Feb 15 02:04:32 tcppdbs005 Had[8970]: [ID 702911 daemon.notice] VCS ERROR
    V-16-.
    Feb 15 02:07:23 tcppdbs005 vxdmp: [ID 736771 kern.notice] NOTICE: VxVM vxdmp
    V-0
    Feb 15 02:07:23 tcppdbs005 vxdmp: [ID 899070 kern.notice] NOTICE: VxVM vxdmp
    V-0
    Feb 15 02:07:23 tcppdbs005 vxdmp: [ID 736771 kern.notice] NOTICE: VxVM vxdmp
    V-8
    Feb 15 02:07:23 tcppdbs005 vxdmp: [ID 899070 kern.notice] NOTICE: VxVM vxdmp
    V-8
    Feb 15 02:07:23 tcppdbs005 vxdmp: [ID 736771 kern.notice] NOTICE: VxVM vxdmp
    V-0
    :TCPPDBS005,TCPPDBS007,TCPADBS023CTCPADBS025H


+ Reply to Thread