UCS Chassis 5108 both IOMs inaccessible

Dear forum users,
Problem description: Sometime our UCS Chassis 5108 and both IOM are suddenly INACCESSIBLE. This happens frequently (Happend 5 times in just a month)
Temporary Solution: To unplug and re-plug Chassis power cords twice, then the Chassis back to normal
Once the Chassis is inaccessible state, below are what we witness on the Chassis:
- All the blades LEDs on chassis-1 shows amber lit
- 8 ports on IOM A and IOM B shows NO LED lit (Not even amber lit)
- Sudden increased in Fan speed on chassis-1
- FI ports which connected to Chassis-1, shows NO LED it
- Please find attached (Chassis-1_Fault.jpg) Fault message on chassis-1
We had tried the following:
- Tried re-sit the uplink cable
- Tried replacing the uplink cable, the port is still down
- Reboot FI
- Decommission the chassis-1
- Recommission the chassis-1
Below is our setup environment:
- 2x UCS 6248UP
- 2x UCS 5108
- 4x UCS 2208XP (2 IOM per chassis)
- 2x UCS B200 M2
- 5x UCS B250 M2
- Firmware 2.0(3b)
- Each IOM has 4 Uplink cables connect to FI
- Chassis Discovery Policy
1. Action (4 Links)
2. Link Grouping Preference (Port Channel)
We hope, someone should have experience with the same problem or have some advise to be shared.
Please let me know should you need any further details of the problem or other information.
Look forward to a positive reply.

Hello Mohamed,
Please open a TAC service request with UCSM and Chassis tech support log files.
Also, include approximate date & time of the event.
Padma

Similar Messages

  • QoS on the UCS Chassis

    I bought a UCS chassis with two blades that are 'qualified' for Cisco Voice products (cm, ccx, etc)....     i decided to not buy the $20,000 worth of VMWARE enterprise pro licenses and the v1000 switch... since the ROI is simply not there (I'd rather have 4 separate pieces of hardware then spend that much money on software)...  anyway...   so now my questions are:
    1. What QoS options are available (if any) on the uplink from the chassis to the 6120s and from the 6120s to my 4500 switch.
    2.  What Qos options are available (if any) on individual hosts inside my vmware.
    3. Can I buy the v1000 switch (which i believe is a few grand) and get some QoS without buying the stupid expensive vmware enterprise pro?
    ...or do i just return the chassis?
    Thanks!

    Terry,
    1. UCS traffic will maintain is CoS markings on the egress of the UCS system.  It's then up to the upstream switch to honor these marking accordingly.  All CoS markings between the Chassis and FI will be honored on the IOM to prioritze traffic accoridngly in the event of congestion using PFC.  Have a read on tha attached paper about UCS & QoS, should help you get an understand on the system handling of QoS/CoS. 
    2. VMware 5 does offer some new auto quality of service options.  Have a read through:
    What's new in vSphere 5 Networking
    3.  No.  Any Distributed Switch (VMware vDS or Cisco 1000v DVS) requires the Enterprise Plus license.  This includes both the Cisco 1000v and VN-Link in Hardware (PTS with Palo M81KR adaptor).
    Regards,
    Robert

  • UCS Chassis I/O problems

    :                         Anyone  else having Cisco UCS 6140/5108 chassis problems that cause the fans to  act up  and the I/O card to have power and temp alarms ver1.3.0 even  when the environmental conditions are OK?

    FYI there are a couple of fixes in this release for these problems.   I believe the problem you are describing is related to this, which is part of the release discussed.
    http://tools.cisco.com/Support/BugToolKit/search/getBugDetails.do?method=fetchBugDetails&bugId=CSCtl43716

  • Ucs chassis qos support

    I  bought a UCS chassis with two blades that are 'qualified' for Cisco  Voice products (cm, ccx, etc)....     i decided to not buy the $20,000  worth of VMWARE enterprise pro licenses and the v1000 switch... since  the ROI is simply not there (I'd rather have 4 separate pieces of  hardware then spend that much money on software)...  anyway...   so now  my questions are:
    1. What QoS options are available (if any) on the uplink from the chassis to the 6120s and from the 6120s to my 4500 switch.
    2.  What Qos options are available (if any) on individual hosts inside my vmware.
    3.  Can I buy the v1000 switch (which i believe is a few grand) and get  some QoS without buying the stupid expensive vmware enterprise pro?
    ...or do i just return the chassis?
    Thanks!

    Terry,
    UC on UCS deployment design considered the hardware features and UC software requirements
    and concluded that there will be no congestion for the IP traffic until it leaves Fabric Interconnect.
    So for the following traffic flow, we do not need additional QoS ( L2 COS ) configuration.
    UC App > vSwitch > physical adapter > IOM > FI > upstream switch
    However, if you have other virtual machines in the blade that contends for the bandwidth, then you environment might need Nexus 1000v or QoS configuration on the UCS.
    Please refer following doc for additional information
    http://docwiki.cisco.com/wiki/QoS_Design_Considerations_for_Virtual_UC_with_UCS#QoS_Design_Considerations_for_VMs_with_Cisco_UCS_B-Series_Blade_Servers
    HTH
    Padma

  • UCS chassis PSU error

    Hi team ,
    We are getting error on PSU for Cisco UCS chassis.
    Its showing NA under performance & voltage tab.
    We tried resetating PSU as well as removing power cable & connecting it back.
    PSU is power-on & operational.
    Can anyone suggest why this issue is & how to get rid of this.

    Hi Ashish ,
    It is possible that you are hitting I2C issues. I recommend to open a TAC case to have an engineer look at your environment and confirm that the I2C bus is congested.
    Here are your options to clear the i2c bus:
    1.You can reseat the IOMs you will need to reseat one IOM at a time starting with the IOM connected to the subordinated fabric,   This should not cause a complete outage.  However, you should perform this during a maintenance window.  Please note, this method does not always clear the i2c bus 100% of the time.  You may have to use option 2 if this is the case. 
    2.The second option is to power down then power back up the chassis.  This method clears the i2c bus 100% of the time.  However, the down side is it is service impacting and will need to be done during a maintenance window.

  • UCS Chassis Not Recognized

    Hello,
    I am having a strange issue with a new UCS environment that we just brought online. One of the four UCS chassis had an issue with its firmware upgrade, specifically the IO module. All of the other componants on all of the other UCS chassis upgraded just fine but these two IO modules hung during the firmware upgrade. I tried to downgrade the firmware on all of the other componants to match the hung IO modules to see if they would come back online and they did not. I also moved the IO modules into a known working chassis and they worked after an auto-upgrade of the firmware! I then swapped them back to their original chassis thinking that they would be ok but they went down again, stating a communication problem and the 6140 reported them as being link-down. I restarted the entire chassis and also swapped the uplink cables and still they show offline. As a last ditch effort I deleted the entire chassis and went back to re-add it and now it will not even detect the chassis.
    I know that the IO modules are fine because I can put them in another chassis and they will come up but as soon I put them back in their original chassis, the 6140's report the link as being down and it will not recognize the chassis. Does this sound like a hardware issue with the chassis itself? Any ideas?

    Can you try using the cables from your "good" Chassis on the failing one?
    I'd suggest you completely power down the bad chassis.  Take the cables from a good chassis, move them over, then power the bad Chassis back up.
    If this doesn't work let me know.  Also ensure your Chassis discovery policy is set to 1 link for now.
    Robert

  • New Chassis with 2204XP IOM & B200 M3 Servers

    Hi,
    We are going to install 2 more additional chassis in existing setup, below is the list of existing setup and new chassis details
    Existing Setup Details:
    2 x 6120XP Fabric Interconnect with 2.0(1t) firmware
    2 x 5108 Chassis with 2 x 2201XP IOM and 8 x B200 M2 servers per chassis
    New Chassis Details:
    2 x 5108 Chassis with 2 x 2204XP IOM and 8 x B200 M3 servers per chassis
    Just wanted to know that is there any issue / problem if we install different configuration IOM and Blade servers in existing setup? If answer is yes then why? And if answer is no then too why? Just want to understand how the things works if answer is no
    -Amit

    Hello Amit,
    B200 M3 servers require a minimum UCS version of 2.0(2).  Please upgrade prior to installing the new blades.
    No problem mixing the newer & older IOMs on the same FI as they fully backwards compatible.  Since the FIs are the 6100 model, the 2204 will not form port-channels on the fabric ports (IOM to FI).  A 6200 would be required for that functionality.
    The webinar given in February covers the hardware & software configuration differences for the new FIs, IOMs, mezz adapters.
    https://supportforums.cisco.com/community/netpro/data-center/unified-computing?view=video
    Matthew

  • UCS chassis uplinks recommendation

    Hi All,
    We currently have a UCS B chassis with 4 blades & 4 uplinks to the FI 6124 (2 from each FEX). We are now looking to populate the chassis with 4 additional blades & was wondering if there is a best practise/guidance document on how many uplinks one should have with a fully populated chassis.
    Thanks
    Kassim

    Kassim
    It all depends on the environment and what degree of oversubscription are you comfortable with between the IOM and the FI.
    As you are already running UCS, I would start by looking at the current utilization to begin with.
    The 2104 is characterized by 8 downlinks (server facing or HIFs) and up to 4 uplinks (FI facing or NIFs).
    A starting config can be either 1, 2 or 4 links i.e 2^x number of links for static pinning.
    Within UCS QoS can be configured so you can specify min bandwith guarantee at the time of congestion.
    If you just want to get a feel for whats out there, most of the time I have come across 2 links per side.
    Again, what I have seen most and not a best practise/guidance by any stretch.
    Thanks
    --Manish

  • UCS - Disjoined L2 - Security

    I havent seen much documents on Disjioned L2 UCS deployment (on security). We have a UCS chassis (5108) and have few B230s on it.. We want to use the same chassis both for internal LAN and DMZ.
    Obviously the first question that comes in is security -  We know that UCS in EHM doesnt act like a switch (with regards to unicasts & broadcasts), but is the design secure enough - logically and physically ? Has anyone come across any security limiations with disjoined L2 ?
    Our thinking was -
    1) Isolation is anyway done on link going from FI to upstream switches - internal LAN VNICs go through a different PINNED uplink than DMZ
    2) Should we consider seperate blades for DMZ ? or running both DMZ and internal on the same blade is fine (with different Vswitches) ?
    3) How about the links going from 2204 FEX to FI ? I know that the VNICs are built upon automatically generated port channels - but is it possible to use 2 different sets of links for internal and DMZ ?
    Regards
    Raj

    I guess it all depends on the organizations security posture. I am finding gov't institutions slow to adopt the segregation mechanisms that UCS and the Nexus products provides, mostly a result of their lack of understanding of the technology and how it's implemented.
    Some people have issues with “mixing” data in the FIs from the different zones. On the other hand, financial institutions seem to be more willing/understanding of the technology in question. But we still see clients requesting physical hardware separation in the DMZ.
    The current implementation of the L2 disjointed works quite well but requires some planning when configuring your vNICs. As you know by default, all VLANs are accessible by all ports. Disjointed L2 is similar to “switchport trunk allow” in the switching world.

  • PSU showing Power save and overall status : N/A

    Hi,
    In UCS chassis 5108,we noticed a 2 PSU was faulty to check properly  we swapped a cable and one of the PSU came up but another PSU was on same status. so we raised PSU and replaced it, but the overall status of the PSU was N/A and its not glowing green LED. Its blinking the Green LED but not continuous glow.
    Please find the attached snap.
    ./Anandraj

    Hi Anandraj,
    What´s your configuration on power groups and firmware version? N+1 or Grid?1.x or 2.x?
    The Blinking green led and status power-save,  I belived this one is a standby.
    When you configure N+1, two PSU work and one Standby mode then If you configure Grid all PSU work together and all Green LED  are glowing.
    Follow below documantion with Power Group configuration:
    http://www.cisco.com/en/US/docs/unified_computing/ucs/sw/gui/config/guide/141/UCSM_GUI_Configuration_Guide_141_chapter29.html#concept_B17F97D8074A4CB590309E3E8C8DD9F9
    PS:
    I found this information below, too.
    The power supply is providing no power to the system but can be powered up
    quickly in the event of another power supply failing. This mode saves power
    in two ways:
    1. It reduces the power consumption of the PSU, as there is no output the PSU
    only requires a small amount of power to keep the capacitors charged and
    circuitry active.
    2. It increases the load on the remaining PSUs shifting them further on their
    efficiency curves.
    Kind Regards,
    Marcus GOmes

  • Does UCS Mini Support 2 Chassis configuration?

    Hello, I have found into the Data Sheet that the UCS 6324 supports until 15 servers, 8 into the chassis and 7 direcct attached UCS C series. 
    http://www.cisco.com/c/en/us/products/collateral/servers-unified-computing/ucs-6300-series-fabric-interconnects/datasheet-c78-732207.html
    the Cisco UCS 6324 Fabric Interconnect embeds the connectivity within the Cisco UCS 5108 Blade Server Chassis to provide a smaller domain of up to 15 servers (8 blade servers and up to 7 direct-connect rack servers).
    But I have heards from a Cisco presentation that the UCS Mini can support till 2 chassis 5108 (one with the FI 6324 and the other one with FEX 2208XP, 2204XP, or 2104XP Fabric Extenders. But I cannot find documentation related.
    Could you please confirm this point? There is any documentation?
    Thanks a lot and best regards,
    Raquel
    I

    True ! FCS + 3 months ?
    However, I've seen many customers with classical UCS configuration and 2 chassis; Cisco offered and offers very attractive bundles, which makes such a low size entry attractive.

  • UCS C series server whether accessible from Linux prompt

    Hi UCS Team,
    I have UCS c series server.
    I am able to check the Fan, power,cooling and equipment and hardware status using the CIMC.
    In CIMC I configured the SNMP and check it using MG soft Mib browser tool.
    The CIMC query the BIOS for getting the information.
    Similarly is there any process to get the BIOS information from the Linux.
    Correct me If I am wrong anywhere.
    My requirement is to query the UCS C server BIOS information using Linux-SNMP without CIMC.(direct access of BIOS from Linux)
    Can I install a Linux supported SNMP client and supported MIBS to query the BIOS level information of UCS using SNMP MG soft Mib browser?

    Hi Keny,
    I tried to powercycle both servers (c260M2 and C460M2) but still no discovered server and also no logg.
    I checked
    Rack Server Discovery Policy == immediate
    Rack management Connecton Policy == Auto Achnowledged
    Chassis/Fex Doscovery Policy == 2 Link and Port-Channel (As I installed 2232PP I moved to Port Channel)
    I am running UCSM Version 2.1(1e)
    FI 6248 Version 5.0(3)N2(2.11e)
    2232PP run on same as FI
    I also have a connected 5108 with IOM 2104 running on 2.1(1e) Version.
    CIMC Version of C260M2 and C460M2 will be checked tomorrow when my kolleague is in the datacenter.
    Server Connection is not really the same like in http://www.cisco.com/en/US/docs/unified_computing/ucs/c-series_integration/ucsm2.1/b_UCSM2-1_C-Integration_chapter_011.html
    Because I still have no VIC Card but the connections are the same.
    C260 GE LOM1 <-> FEX2 Port 11
    C260 GE LOM2 <-> FEX3 Port 11
    C260 10 GE LOM1 (PCIE Slot4) <-> FEX2 Port 12
    C260 10 GE LOM2 (PCIE Slot4) <-> FEX3 Port 12
    GE LOM ist position 13 and 10 GE LOM is position 11 in the techspech of c260
    http://www.cisco.com/en/US/docs/unified_computing/ucs/c/hw/C260/install/overview.html#wp1243740
    C460 GE LOM1 <-> FEX2 Port 10
    C460 GE LOM2 <-> FEX3 Port 10
    C460 10 GE LOM1 <-> FEX2 Port 9
    C460 10 GE LOM2  <-> FEX3 Port 9
    GE LOM ist postion 8 and 10 GE LOM ist position 7
    http://www.cisco.com/en/US/docs/unified_computing/ucs/c/hw/C460/install/overview.html#wp1243740
    With C260 I also tried a reboot with disconnected 10 GE Ports, but also no discoverd Server until now.
    Regards Alois

  • UCS C-Series Server discovery with UCSM integration failed

    Hi, i try to install C460M2 and a C260M2 via fex 2232PP
    Because we are waiting for delivery of VIC Cards I have to install the Servers via 2 wire.
    This means I connected each Server with the 2 LOM gigabit ports to the nexus 2232PP (one port ono each nexus), and I also connected the 10 ge LOM to each of the nexus 2232PP.
    My problem is the nexus 2232 are discovered, but no one of my servers get discovered. All Ports have link up, I can not find any log message on discovery error.
    I know the 10 GE Server Ports are not supported, but I think the servers should get discovered.
    Does anybody have an idea, wherecan I find all logg information and what filter ist suggested ?
    Best Regards Alois
    Sent from Cisco Technical Support iPad App

    Hi Keny,
    I tried to powercycle both servers (c260M2 and C460M2) but still no discovered server and also no logg.
    I checked
    Rack Server Discovery Policy == immediate
    Rack management Connecton Policy == Auto Achnowledged
    Chassis/Fex Doscovery Policy == 2 Link and Port-Channel (As I installed 2232PP I moved to Port Channel)
    I am running UCSM Version 2.1(1e)
    FI 6248 Version 5.0(3)N2(2.11e)
    2232PP run on same as FI
    I also have a connected 5108 with IOM 2104 running on 2.1(1e) Version.
    CIMC Version of C260M2 and C460M2 will be checked tomorrow when my kolleague is in the datacenter.
    Server Connection is not really the same like in http://www.cisco.com/en/US/docs/unified_computing/ucs/c-series_integration/ucsm2.1/b_UCSM2-1_C-Integration_chapter_011.html
    Because I still have no VIC Card but the connections are the same.
    C260 GE LOM1 <-> FEX2 Port 11
    C260 GE LOM2 <-> FEX3 Port 11
    C260 10 GE LOM1 (PCIE Slot4) <-> FEX2 Port 12
    C260 10 GE LOM2 (PCIE Slot4) <-> FEX3 Port 12
    GE LOM ist position 13 and 10 GE LOM is position 11 in the techspech of c260
    http://www.cisco.com/en/US/docs/unified_computing/ucs/c/hw/C260/install/overview.html#wp1243740
    C460 GE LOM1 <-> FEX2 Port 10
    C460 GE LOM2 <-> FEX3 Port 10
    C460 10 GE LOM1 <-> FEX2 Port 9
    C460 10 GE LOM2  <-> FEX3 Port 9
    GE LOM ist postion 8 and 10 GE LOM ist position 7
    http://www.cisco.com/en/US/docs/unified_computing/ucs/c/hw/C460/install/overview.html#wp1243740
    With C260 I also tried a reboot with disconnected 10 GE Ports, but also no discoverd Server until now.
    Regards Alois

  • FCoE options for Cisco UCS and Compellent SAN

    Hi,
    We have a Dell Compellent SAN storage with iSCSI and FCoE module in pre-production environment.
    It is connected to new Cisco UCS infrastructure (5108 Chassis with 2208IOM + B200 M2 Blades + 6248 Fabric Interconnect) via 10G iSCSI module (FCoE module isn't being used at th is moment).
    I reviewed compatibility matrix on interconnect but Compellent (Dell) SAN is only supported on FI NXOS 1.3(1), 1.4(1) without using 6248 and 2208 IOM which is what we have. I'm sure some of you have similar hardware configuration as ours and I'd like to see if there's any supportive Cisco FC/FCoE deployment option for the Compellent. We're pretty tight on budget at this moment so purchasing couple of Nexus 5K switches or something equipvalent for such a small number of chassis (only only have one) is not a preferred option. If additional hardware acquisition is inevitable, what would be the most cost effective solution to be able to support FCoE implementation?
    Thank you in advance for your help on this.

    Unfortunatly there isn't really one - with direct attach storage there is still the requirement that an upstream MDS/N5k pushes the zoning to it.  Without a MDS to push the zoning the system it's recommended for production.
    http://www.cisco.com/en/US/docs/unified_computing/ucs/sw/gui/config/guide/2.0/b_UCSM_GUI_Configuration_Guide_2_0_chapter_0101.html#concept_05717B723C2746E1A9F6AB3A3FFA2C72
    Even if you had a MDS/N5K the 6248/2208's wouldn't support the Compellent SAN - see note 9.
    http://www.cisco.com/en/US/docs/switches/datacenter/mds9000/interoperability/matrix/Matrix8.pdf
    That's not to say that it won't work, it's just that we haven't tested it and don't know what it will do and thus TAC cannot troubleshoot SAN errors on the UCS.
    On the plus side iSCSI if setup correctly can be very solid and can give you a great amount of throughput - just make sure to configure the QoS correctly and if you need more throughput then just add some additional links

  • Network link LED on UCS blades

    Hi,
    we got a working UCS system (6248 FI , 5108 chassis with 2204 IOM and B200-M3 blades) all is working fine but the network link LED are off. Is this a normal behaviour ? Since according to documentation they should blink green.
    regards

    Hi Lukasz,
    I'm assuming we are speaking about the network LED 's on the IOM which are off.
    What about the LED's on the respective ports on the FI?
    What's the color of the LED's on the power supplies?
    Thanks,
    Majid

Maybe you are looking for