XenServer 6.1 SAN boot on B200M2
I'm new to XenServer. We have UCS with B200M2's with the M81KR CNA in them. We primarily use VMware ESXi and we SAN boot all of those machines. We have had no issue SAN booting ESXi servers. I've created a new serivce profile template for the XenServer's and brought up a couple machines. I've allocated the storage and when I got to install XenServer it doesn't see the boot LUNs. I know the storage is allocated properly. I can see the LUNs when the blade is booting and I can install ESXi onto the boot LUNs without issue. The issue is with XenServer seeing the LUN. I did the F2 > mulitpath option that it says to do in the Appendix of the install guide for XenServer but that makes no difference. What needs to be done to have XenServer see a boot LUN on a B200M2?
6.0
Please see the UCS Support Matrix tool for further info:
http://www.cisco.com/web/techdoc/ucs/interoperability/matrix/matrix.html
Support is coming in the upcoming UCSM 2.1 release before the end of the year (tentatively). I'm just confirming if support will be included in a future 2.0.x patch or not.
Regards,
Robert
Similar Messages
-
UCS 1.4(3M) BIOS Issue ? no network/san boot when diskless
Recreated faithfully on multiple blades and chassis with VIC cards. Have many SAN boot nodes in production and testing and after upgrading firmware this week anytime I apply my store policy of "diskless" to a blade I lose the ability to boot from network. After much effort to work around comes down to no matter what I do with diskless storeage policy, the configured boot order will never become the actual boot order. In the BIOS all network boot options/reference are invisible, cannot enable nor add to options. Upon removing diskless storeage profile and applying "any other that I have" I see in FST that it's powering up internal storage, starting up and updating pre-boot environments, and other things related that connected the dots. Returning to BIOS all reference and options have returned as I expected due to actual boot order screen updating to match configured order and hba's of course login the fabric right away, boot is fine. While I can faithfully replice across many, many blades, they are all the models with same mezz running 1.4.3 so curious if anyone else is seeing, particuaraly now with new version out, as I have done last few weeks without issues leading me towards the firmware.
In a nut shell I see that when going diskless this is actually shutting down the storage/raid controller. i can see that clearly but no bios access to it and not sure what else is or why it's affecting network and hba also since I cannot see, it's shutting more down than low powering drive side raid. I use different profiles due to some blades having disks removed, some retain them. moving a san boot service-profile dynamically around multiple chassis will fail to associate with errors on some due to disk setup and all that. some i have disks pulled and diskless based on documentation and best practices as this customer needed several san boot only blades in that pool. But yes if I use a storage profile allowing "any disk config" for association whether it has disks, no disks, raided drives, replaces the diskless policy in same service profile it faithfully follows my boot order and boots via hba's across multiple blades, up to about 8 or 10 as of today.....b200-m2's with vic's
thx
daveMy UCS manager firmware and Adapter is 1.3.1p,
the boot policy has been set to Boot to SAN,
I checked "Hardware and Software Interoperability Matrix for UCS System Release 1.3.1"
It seems the 1.3.1 only can support the "windows 2008 r2 x64 w/Hyper-v",
so, I updated the UCS manager Bios and the adpter firmware to 2.0, but the I/O module still is 1.3.1p
I can install the "windows 2008 r2 x64" successfuly...
Now, my concern is I have a mix component software versions on my ucs blade(I used the UCS manager 1.3.1, IO module 1.3.1, the UCS Bois 2.0.1 and the adapter 2.0.1)
will the mix component software versions cause some other issues? Dose anyone can give me some suggestion?
Thanks a lot -
During Solaris 10 Installation on a SAN Based disk, SUN Fire V10 was not rebooting after the first CD installation. It does memory dump...
I really don't like san boot for operating system. You have too many points of failure for the basic operation.
Anyway, during installation, ensure that you access the disk using just one path (using zoning for example). After you enable and configure MPxIO, you could enable other paths. -
SAN boot disk in a cluster node?
As far as I remember, SC requires local boot disk.
Is it possible to use the SAN boot disk? Can this configuration work being just not supported by SUN or there is some technical limitation ?
I am thinking of low price configuration with two diskless X2100s + HBAs and SAN storage. Possible?
thanks
-- leonAs far as I remember, SC requires local boot disk.
Is it possible to use the SAN boot disk? Can this
configuration work being just not supported by SUN or
there is some technical limitation ?The rule for boot disks gos like this:
Any local storage device, supported by the base platform as a boot device, can be used as a boot device for the server in the cluster as well. A shared storage device cannot be used as a boot device for a server in a cluster. It is recommended to mirror the root disk. Multipathed boot is supported with Sun Cluster when the drivers associated with SAN 4.3 (or later) are used in conjunction with an appropriate storage device (i.e. the local disks on a SF v880 or a SAN connected fiber storage device).
So your boot disk can be in a SAN as long the base platform supports it as a boot disk and it is not configured as a shared lun in the SAN (i.e. visible to other nodes than the one that uses it as boot disk).
I am thinking of low price configuration with two
diskless X2100s + HBAs and SAN storage. Possible?You need to check the support matrix of the storage device you plan to use, if it is supported as a boot device for the X2100 + HBA. If the answer is yes, you just must make sure that this lun is only visible to that X2100.
Greets
Thorsten -
B230-M2 - MK81KR - Win 08 R2 San Boot
Chaps.
I need to SAN boot this server with Win 08 R2 on the metal.
Service profile works with VMware, so the Zoning/Masking is correct.
On install from 08 Disk, no Hard Drives shown, I loaded the Cisco Storage Driver, now I see the Lun twice, Disk1
Unpartitioned space and Disk2 Unpartitioned space.
There is a Bios not enabled warning when I try to install.
UCS Version 1.4.3Q
Driver Disk : 1.4.1
Please assist
DanieNo change on either option.
On the CD sequence.
I clicked load driver, mounted the 1.4.3 Drivers cd, browsed to Windows\Storage\Cisco\MK81KR\Win08R2\64
I listed the driver.
I then mounted the Win08 cd, and clicked next to load the driver, it could not find the path, as I had un-mounted the driver disk, so remounted 1.4.3, clicked next, green bas flashed a couple of times, and I got to the Where do you want to install screen. My 40GB disk showed as offline.
Mounted the Win08 Disc, and clicked refresh, twice....
Disk stayed offline, show details on the error allowed me to bring the disk online.
Now the error stated that the BIOS is not setup for booting of this controller, ignored the warning, and clicked next.
Setup was unable to create a new system partition or locate an existing system partition.
I can create and delete partitions manually.
EFI disabled and the same install sequence, same results
Thanx -
San Boot Win2K8 R2 SP1 x64 on UCS blades B with M81KR installation issue
I am trying to install a storage San Boot Win2K8 R2 SP1 x64 on UCS blades,(There are only 1 path to the OS installation drive during the installation.)
I loaded drivers for the M81K2 following the "UCS Windows Install Guide" , and I can see the LUN I mapped from XIV. I was able to create and format partitions, make it online, but when
I try to install, windows gives the error:"Windows cannot be installed to this disk. The computer's hardware may not support booting to this disk. Ensure that the disk's controller is enabled in the computer's BIOS menu."
any one know this issue?
ThanksMy UCS manager firmware and Adapter is 1.3.1p,
the boot policy has been set to Boot to SAN,
I checked "Hardware and Software Interoperability Matrix for UCS System Release 1.3.1"
It seems the 1.3.1 only can support the "windows 2008 r2 x64 w/Hyper-v",
so, I updated the UCS manager Bios and the adpter firmware to 2.0, but the I/O module still is 1.3.1p
I can install the "windows 2008 r2 x64" successfuly...
Now, my concern is I have a mix component software versions on my ucs blade(I used the UCS manager 1.3.1, IO module 1.3.1, the UCS Bois 2.0.1 and the adapter 2.0.1)
will the mix component software versions cause some other issues? Dose anyone can give me some suggestion?
Thanks a lot -
I have just created an ABE on a SAN device, upgraded it and then booted from it.
Its been a long journey, and involved a SR with oracle.
I was new to it and am astonished at how hard it has been. Not hard in the sense that anything was over complicated, but the tangle mess of patch levels and gathering information on server firmware, SAN boot, HBA firmware, patches etc has been a nightmare. Not sure whether its me just new to patching solaris 10 etc, but I have literally trawled through a couple of dozen documents, countless web browsing and a few grey hairs just to achieve this.
I will hold my hand up, some of it was of our own making - I was patching a system that had not been touched in a number of years.
Anyone had a similar experience or does it get easier?!?!?I really don't like san boot for operating system. You have too many points of failure for the basic operation.
Anyway, during installation, ensure that you access the disk using just one path (using zoning for example). After you enable and configure MPxIO, you could enable other paths. -
If i have a 3rd domain on a m5k. and it is supposed to boot from SAN. how can I go about doing it?
If thats the case like if my storage got 2 controllers and each has a fc path to the domain 3,
wun there be 2 luns seem on system side?
how would I be able to do the stmsboot? Even if i do a stmsboot after OS, won't the system be unable to boot after?.
or do I map 2 individual luns to server, then raid @ server side? or do we only do mapping on switch side?
Thank you all for your kind attention.Hi Tejas,
We are going to need some further details to help narrow down the problem.
1) Was this Boot from SAN working previously for the ESX blade, you then upgraded to 2.0(1s) and are now having the issue?
or
Is this a new Boot from SAN setup on 2.0(1s)?
2) You mentioned that you can see the WWPN of your blade in your Fiber Channel switch. Have you verified that the WWPN of the EMC VNX is also visible in your Fiber Channel switch?
3) Have you checked your zoning configuration on the Fiber Channel switch, ensuring that the WWPN of your blade is permitted to communicate with the EMC VNX storage array?
4) Have you also checked your LUN masking on the EMC storage array? Ensure that you have configured the storage group to allow the WWPN access to the LUN
5) Depending on if the array is active/active or active/passive, it is also important to verify which storage processor currently owns the LUN and ensure that the vHBA communication is hitting the appropriate storage processor
6) Generally most operating systems will boot from LUN ID 0. So you can compare the presented LUN ID in EMC to what you configured in your UCSM Boot policy
7) Also ensure that you have correctly configured the order of your Boot from SAN targets (primary vHBA - primary target, secondary vHBA - primary target, etc...)
Other tips available via the following:
Troubleshooting SAN Boot and SAN Connectivity Issues
Hope that helps. Let me know how you go.
Cheers,
Michael -
I would like some clarification regarding the SAN Boot policy. I was always under the impression you were suppose to use the WWPN of the storage port. But what if you have many storage ports coming from the array (16 for example), this would not be possible.
Is the purpose of this policy to simply force the HBA to login into the SAN fabric?
If so, I can use one from my actual array or just make one up, correct?
Since all the zoning is being done by the SAN switch (Brocade/MDS/Nexus) and LUN masking at the array level, I would think putting the actual WWPN of a storage port is not necessary.
Thanks for any clarification.My thought was you could just throw a valid wwpn in (any valid wwpn, doesn't have to be the array) again just to log in to the fabric. The it would scan the bus and off you go.
When I tried that with our try and buy setup and a NetApp array, doing windows 2008 r2 san boot, I could never get the luns to show up. We put in the arrays 4 wwpns and that got it to show up.
im not sure why it matters, and I think we had something else wrong. if you look at blogs online almost all of them show to just throw a wwpn in there to start flogi and that's it.
If you have 16 wwpns that could be an issue. As you probably know the boot policy only lets you put a primary and secondary for each HBA. I would try placing a random wwpn in and see if it works. -
SAN Boot Target-Cisco folks, please advise
Hi:
Just want to know the difference of the following 2 configurations for SAN Boot for UCS with 2 WWPNs being presented from SAN Storage
Array: I know 1 will work, but how about Option 2????
Cisco UCS manual DOES NOT go into the details needed.
1.
SAN Primary, SAN Target Primary- vHBA-A - Fabric A
SAN Secondary, SAN Target Primary - vHBA-B - Fabric B
VS
2.
SAN Primary, SAN Target Primary - vHBA-A - Fabric A
SAN Secondary, SAN Target Secondary - vHBA-B - Fabric B
Thanks.
KP SimI don't think the system will let you introduce a secondary target on the second vHBA if you haven't defined a primary target for that vHBA.
SAN boot works as follows if you have a policy with 4 targets and 2 vHBAs:
1. try primary vHBA, first (primary) target
2. if that doesn't work, try secondary target on primary vHBA
3. if that doesn't work, switch to secondary vHBA, primary target
4. if that doesn't work, try secondary target on secondary vHBA -
Hi All,
How to configure SAN Boot on the UCS C Series platform ?
On B Series, we can configure the boot order from vHBA, but on the UCS C Series I found nothing.
The C Series SAN boot had being enabled and also the boot target was configured on the vHBA properties, but the C Series can't boot from SAN
Attached are the detailed of the configuration on C Series
Please help, if you have C Series boot from SAN succesfully
Thanks,
GofiHi Padma,
Yes, that will be the flogi into the upstream switch.
Below are the configurations of Nexus 5K, while I used 2 N5k as the upstream FCoE switch
N5K1# sh run int vfc113
!Command: show running-config interface vfc113
!Time: Wed Jun 17 03:21:56 2009
version 5.2(1)N1(5)
interface vfc113
bind interface Ethernet113/1/1
switchport trunk allowed vsan 10
no shutdown
N5K1# sh run int eth113/1/1
!Command: show running-config interface Ethernet113/1/1
!Time: Wed Jun 17 03:22:05 2009
version 5.2(1)N1(5)
interface Ethernet113/1/1
switchport mode trunk
switchport trunk allowed vlan 10,120,125,130,135,200-201
N5K1# sh run | sec fex
feature fex
fex 113
pinning max-links 1
description "FEX0113"
fcoe
fex 123
pinning max-links 1
description "FEX0123"
N5K1# sh feature | i fcoe
fcoe 1 enabled
fcoe-npv 1 disabled
N5K2# sh run int vfc123
!Command: show running-config interface vfc123
!Time: Wed Jun 17 03:24:12 2009
version 5.2(1)N1(5)
interface vfc123
bind interface Ethernet123/1/1
switchport trunk allowed vsan 20
no shutdown
N5K2# sh run int eth123/1/1
!Command: show running-config interface Ethernet123/1/1
!Time: Wed Jun 17 03:24:29 2009
version 5.2(1)N1(5)
interface Ethernet123/1/1
switchport mode trunk
switchport trunk allowed vlan 20,120,125,130,135,200-201
N5K2# sh run | sec fex
feature fex
fex 113
pinning max-links 1
description "FEX0113"
fex 123
pinning max-links 1
description "FEX0123"
fcoe
N5K2# sh feature | i fcoe
fcoe 1 enabled
fcoe-npv 1 disabled
The CNA server connected to FEX Nexus 2K, while the Nexus uplinks are controlled by 2 Nexus 5Ks (vPC on the N5K)
Any suggestions ?
Thanks,
Gofi -
UCS Central - Making Boot Policy - SAN boot target Provisioning
Gents
Some objects is always related to the domains and need to be provisioned in global like
VLANs: but we can did it by Provisioning VLANs with UCS Central Using VLAN ID Aliasing.
VSANs: the same idea in VLAN ID Aliasing.
Boot Policy adding SAN boot target : what about this issue how can I provision it so I can configure it globally in abstract way then it will take targets depend on domain during associating the global service profile.@Gmonroy
Thanks for support
But if I need to make a global Service profile temp so I can deploy SF from it in any domain, and if I referenced this service profile temp to one boot policy so i need to unbind each service profile created from this temp to associate a different boot policy in each domain ? " I will lose the benefit of updating service profile template "
Or I have to create different template per domain ? " I will lose the key benefit of centralizing Templates in UCS central"
Or there is any workaround to deply this senario. -
I've been booting my B200-M1 (and M2) blades from the SAN with no issues, but when I try it with the B440, it just hangs. I know it sees the LUN because when I boot from the install ISO I can select the lun as the install location. I'm also using the same boot policy as my B200 blades, so the targets are correct. The firmware on the system is 1.3.1n.
Anyone else having a problem SAN booting a B440?Update and resolution:
Update: there was no problem with CentOS 5.5 or VMware 4.x - I just hadn't checked those when I posted this message.
Update: there is a problem with CentOS (or RHEL) < 5.5 - see fix below
Solution: during the install boot, use the following: 'boot: linux mpath'
The problem was that I wasn't using 'boot: linux mpath' during the install boot of CentOS. By not using mpath, something was messed up with the subsequent SAN boot.
Note: even with the Cisco drivers, this wasn't resolved until I used mpath
Note: despite the fact that Cisco doesn't have a driver for RHEL 5.3, it still works if you use mpath during install
Cheers! -
UCS ESX5.1 with VNX5300 unable to SAN-boot after installation
Hi All,
We have a scenario here, which EMC support claim that they've see this elsewhere.
There is 1 units of CISCO UCS B200 M1 (Palo) with 4 Blades connecting to CISCO Fabric Interconnect 6120XP.
CISCO Fabric Interconnect 612XP has configured with UPLINK to Brocade Switch 5100B.
Meanwhile, VNX5300 is connecting to Brocade Switch 5100B.
Findings:
1. WWN of the Storage and Blades appear on SAN Switches
2. Zoning between 4 Blades and VNX5300 done, initiator of Blades is not appear in UNISPHERE.
3. Have to manually registered initiator for the Blades in the Unisphere
4. Assign LUNS (VNX5300) to the Blades. Blades detected the LUNS.
5. Installed ESX on boot LUNS. But when we wanted to boot up the ESX host, it shows LUN not detected & can’t boot-up the ESX host.
Highlighted issue and concern as below:
1. Why WWN of the Blades not appear on UNISPHERE?
2. Why during ESX boot up, LUN is not detected?
Any suggestion..? Thanks!
Rgds,
Nazrisounds like a zoning issue, WWPNs of vHBAs should show up in the Storage System Connectivity status window in Unisphere. For new hosts in a boot from SAN configuration you have to register them manually because there isn't yet an OS there to auto-register. This can be automated with naviseccli.exe.
Each ESXi host in a boot from SAN configuration will need a unique Storage Group and Boot LUN assigned to the storage group. The boot LUN host ID must match what is configured in UCSM on the boot policy. -
Difficulty with SAN boot of solaris 10 (on V245) from HP EVA 4000 -
Hi
This is my first post. I'm trying to get my head around what's preventing me booting a V245 solaris 10 from a LUN presented from an HP EVA 4000.
The machine boots up ok from a local disk. So, i've done the LUN masking (presentation) on the EVA and the Brocade zoning - solaris can see the disk.
I've formated it close to the current internal root disk, and had used ufsdump/ufsrestore and installboot to copy the contents of / onto the new LUN , which i had mounted as /mnt.
My problem is i just can't seem to get at the new disk at the ok prompt - i'm never quite sure that i'm booting from the correct disk.
I believe all the compatibility with o/s qlc card drivers and the EVA are in order.
The qlc HBA is as follows:
Sun SG-XPCI2FC-QF2/x6768A 1077 2312 1077 10A and have suitable patch 119139-33
Here's some output:
root@bwdnbtsl02# uname -a
SunOS bwdnbtsl02 5.10 Generic_137111-01 sun4u sparc SUNW,Sun-Fire-V245
root@bwdnbtsl02#
root@bwdnbtsl02# luxadm probe
No Network Array enclosures found in /dev/es
Found Fibre Channel device(s):
Node WWN:50001fe15009df00 Device Type:Disk device
Logical Path:/dev/rdsk/c4t600508B4001066A60000C00002B50000d0s2
root@bwdnbtsl02#
Now down at ok prompt, probe-scsi-all can see the EVA (HSV) LUNS
(Here's a sample of the output)
ok probe-scsi-all
/pci@1f,700000/pci@0/SUNW,qlc@2,1
************************* Fabric Attached Devices ************************
Adapter portId - 640100
Device PortId 640000 DeviceId 1 WWPN 210000e08b8a3e4f
Device PortId dc0000 DeviceId 2 WWPN 50001fe15009df09
Lun 0 HP HSV200 6220
Lun 1 DISK HP HSV200 6220
Device PortId d20000 DeviceId 3 WWPN 50001fe150092f19
Lun 0 HP HSV200 6220
/pci@1f,700000/pci@0/SUNW,qlc@2
************************* Fabric Attached Devices ************************
Adapter portId - 630100
Device PortId 630000 DeviceId 1 WWPN 210100e08baa3e4f
Device PortId 780000 DeviceId 2 WWPN 50001fe15009df08
Lun 0 HP HSV200 6220
Lun 1 DISK HP HSV200 6220
Device PortId 6e0000 DeviceId 3 WWPN 50001fe150092f18
Lun 0 HP HSV200 6220
So, i attempt to boot with the following:
ok boot /pci@1f,700000/pci@0/SUNW,qlc@2/fp@0,0/disk@w50001fe15009df08,0:a
Boot device: /pci@1f,700000/pci@0/SUNW,qlc@2/fp@0,0/disk@w50001fe15009df08,0:a File and args:
Can't read disk label.
Can't open disk label package
Can't open boot device
{1} ok
If anyone can put me straight here i'd be most grateful.
thanks johnhi,guys:
i am interesting in your question. i have never try it like your experience.
maybe the cause is drive . either the OS is running . the drive can load. or relation with the style of server .
i think you can try it like the following again.
when you install OS , you chocie EVA disk .
remeber . if the test is complete. you must tell me .
the following is my contact: [email protected]
best regards
Maybe you are looking for
-
Using M1 and F1 ports in OTV set up
I am building a typical OTV architecture (on a stick, not inline) on Nexus 7000s; the join and internal interfaces in my OTV VDC are using M1 card ports. Can the other end of those links -- i.e. the interfaces in my Agg VDC -- use ports on my F1 card
-
We are having this issue across multiple IOS devices ranging from IOS version 5.x up to the latest 6.1.2. Many (not all) emails sent to the devices are unreadable and simply state: "This message has not been downloaded from the server". Microsoft bel
-
Error in mapping of request message (IDoc)
Hi! I am facing with the following mapping error. I try to map data element MT_Material to IDoc MATMAS.MATMAS02 com/sap/xi/tf/_MM_MT_Material_MATMAS02_</Trace> <Trace level="1" type="T">com.sap.aii.utilxi.misc.api.BaseRuntimeException: RuntimeExcepti
-
Is lion compatible with quark express and adobe photoshop?
I have heard that once you upgrade to Lion, Quark Express and Adobe Photshop CS no longer work right. Is this true?
-
Okay, so I am very new at AI and I was working on an assigment for my class when the cursor quit giving me the option to resize anything! I have an assignment due tonight and I'm a little bit upset. Does anybody know what I am talking about and what