vmnic redundancy and UCS

Similar Questions

• Configuration & SRI and UCS description

  Hello

  Operating from my previous post, I would really appreciate if a good person want to throw some light on specifications of SRI and UCS, and configurations usually supproted on them and all the descriptive stuff about the same, I am preparing for these subject (especially the old one, which is completely new to me) and would like some accurate and precise information on the forum!

  Search for SRND - relevant on the product that you want to learn more about.

• What is the difference between redundancy and recovery for retention window?

  Hi all
  
  I'm still trying to get a grip on this concept of planning RMAN backup.
  
  I recently came across a situation like this.
  
  I have backups on a redundancy of 1.
  
  I have instance1, who had a job of GRID to run backups RMAN as follows
  
  Database incremental level = 0 a week at sea at 02:30
  I have backups daily, incremental level = 1 every night
  I have backups of the log of the archives through the day.
  
  What happened, was apparently on a TUE some time, the development team remove data in almost all of the tables in the schema.
  The this didn't discover until noon or so on Wednesday.
  
  I went to try to restore... and RMAN could not do. I had found a SNA for early TUES, they did... but I got the following error:
  
  RMAN > RUN
  2 > {}
  3 > SET up to this THAT SNA 4590383198282.
  4 > RESTORE database;
  5 > RECOVER DATABASE;
  {6 >}
  
  executing command: SET until clause
  
  From restoration to the 28 April 10
  allocated channel: ORA_SBT_TAPE_1
  channel ORA_SBT_TAPE_1: SID = 639 device type = SBT_TAPE
  channel ORA_SBT_TAPE_1: NMO v4.5.0.0
  allocated channel: ORA_DISK_1
  channel ORA_DISK_1: SID = 638 type device = DISK
  
  File number of the data file = 1 name=+DATADG/instance1/datafile/system.284.697137287
  RMAN-00571: ===========================================================
  RMAN-00569: = ERROR MESSAGE STACK FOLLOWS =.
  RMAN-00571: ===========================================================
  RMAN-03002: failure of the command restore at 2010-04-28 15:53
  ORA-01180: cannot create the data file 1
  ORA-01110: data file 1: '+ DATADG/instance1/datafile/system.284.697137287 '.
  
  What I did... of research, it seems that since I don't have a level 0 incremental of before, he could not reset the database in time. Precedent has been removed since, I guess I had a redundancy of 1.
  
  So, I thought that I had a week of backup they way I put it... but that assumes that the errors are found fairly quickly before the next level = 0 backup.
  
  I want at least 1 to 1.5 weeks of worth of backups. I have the problem that we do have many tapes in rotation, so I can't go because too long..may 2 x weeks at the top.
  
  I'm revising my strategy, but I am confused that I read about the difference between retention RMAN setting with window of recovery / redundancy.
  
  Can someone give me an explanation how do you understand? I read on the web and in books to oracle, and for some reason, it is not clear to me what is the best way to go... especially when I see this window of redundancy and recovery can affect?
  
  Thank you in advance,
  
  Cayenne

  Redundancy = 1 means rman will mark any more old than the backup of the last full or incremental backup of level 0 as obsolete. So it is very likely that the backup that you need for this restore has been removed after work Wednesday morning.

  The recovery window will tell rman to keep (or not obsolete brand) all necessary backups to restore to any point in time in the recovery window.

  Bjoern

• ESXi vmnic teaming and multiple access switch

  Hello

  We intend our ESXi to connect to 2 different access switch to increase redundancy. The two switch connect to core via etherchannel. Currently connect us 2 vmnic to 2 buttons (one on one), assign 2 vmnic vswitch as uplink adapter, set these 2 uplinks as active adapters and load balancing game based on the virtual port id.

  Everything seems fine until I found KB 1001938 said "ESX/ESXi host only supports grouping NETWORK adapters on a single physical switch or stacked switches. Is there a problem with our Installer? If so, how should we configure redundancy of the network (different access switch?)?

  dhchentw wrote:

  Everything seems fine until I found KB 1001938 said "ESX/ESXi host only supports grouping NETWORK adapters on a single physical switch or stacked switches. Is there a problem with our Installer?

  There is no problem and your solution is fully supported. The article describes the aggregation of standard links 803.2ad (called "IP Hash" on ESXi) who has the above requirements with a single or bunk physical switches.

  When you use the method of identification of the Port without problem, you can connect multiple physical switches in the host, make sure that all the VLANS are available on all links of the switch.

• Integrating Active Directory and UCS Manager

  I'm looking to create an LDAP authentication provider in the UCS Manager that will authenticate users in Active Directory. I see the configuration guide UCS that a schema change is required to add a new attribute for user accounts and the guide details what the new attribute should be. However there are no detailed instructions on how to make the change to AD. I imagine some sort of import LDIFDE is required, but does anyone have more detailed steps on how to do it?

  Thank you

  You can ssh in your UCS, go to the NxOS prompt and test authentication as follows:

  Laurel - A (nxos) # test cpaggen aaa cisco group ldap
  the user has been authenticated
  Laurel - A (nxos) # test aaa group ldap cpaggen cisco1
  user authentication failed
  Laurel - A (nxos) # test aaa group ldap foo doesntexist
  user authentication failed
  Laurel-a. (nxos) #

  Make sure that this part of work. The role assignment comes from CiscoAVPair and the value must be a shell: roles = 'admin' If you want the user to be an administrator. CiscoAVPair must be an attribute of the user object. I've attached a screenshot of Wireshark for a successful authentication and authorization.

  You will also find the definition of the user and configuration of my UCS.

• ESXi Shared memory of commom and UCS B200M2

  I vSphere 4.x environment to IBM 3650 and half B200M2 (a group consisting of the 3650 and a group consisting of the B200M2).  IBM machines are 2 socket quad core with 34 GB of memory and Cisco are 2 Socket 6 core with 48 GB of memory.  Workloads on IBM and Cisco ESXi hosts are the same.  For most of 2008 and Windows 2003 machines.  On IBM machines, I move 1 GB, 2 GB of shared memory (common shared memory - amount of memory that is shared by all virtual machines powered and vSphere services on host machine) on each ESXi host.  On Cisco machines, I get around 15 MB on each ESXi host. Considering that it is one of the great features of VMware (sharing of memory) why is there such a huge disparity between the machines?  Is it because the Cisco blades NUMA machines?  Is there a setting I'm missing?  It's a huge difference.  I don't regret getting the UCS stuff I like it but it is the big novelty of the VMware that does not work with the UCS.

  Please take a look at the following:

  http://KB.VMware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalID=1020524

  Transparent Page Sharing on the Nehalems won't be out until the memory usage is almost 100%.

  Edit: should read: Transparent Page Sharing on the Nehalem and later generations will not kick until the memory usage approaches 100%.

• VLAN Management with the 1000V and UCS

  If I want to use 10 VLAN management, to configure the following:

  1000V - VM vEthernet port profile coelio with VLAN 10, Ethernet uplink profile includes 10 VLANS

  UCS - vNIC in the Service profile, VLAN 10 and include the UCS uplink trunk VLAN 10

  Switch upstream: include 10 VLANS in trunk port.

  Ok?

  Now, this management VLAN, I then jeutiliser can for everything? I mean to 1000V management VLAN, vSPhere management, management FI and switch? For example management of 1000V or different from others?

  Hi Atle,

  Yes, is that what you mentioned, it's good in terms of activities you need to perform. However, I wish to add a few:

  (1) define him vlan on the UCS - except if you do this, that you won't be able to add to the vnic. More once you have set the vlan automatiaclly will get added to the list of trunk on uplink ports (unless you l2-disjoint set up)

  You can have the same vlan for all management, unless you have some traffic that you would not want some devices to see / meet.

  . / Afonso

• Redundant (and wrong?) ADDED_TO_STAGE events

  I'm always in the middle of this diagnosis, but thought others might have seen, or it may stick to an explanation possible for problems in their application, maybe just with the latest SDK or OS.  I am not sure yet that it's anything other than to hurt me, but of course, it does not match my understanding of how things are meant to work.

  I have an app with a really simple Interface, the relevant material being the root Sprite, a homepage Sprite in that and in this page a Sprite that contains the bitmap I'll call graph.  I am facing a problem at startup when the graph is two ADDED_TO_STAGE events, even if it is NOT added multiple times or removed.

  I put things up by adding an ADDED_TO_STAGE listener to each class constructor.  In the root of the ADDED_TO_STAGE sprite, I build the main page and addChild(), which leads to get an ADDED_TO_STAGE (immediately).  In this handler, I build the graph (and much more) and addChild(), if she then gets an ADDED_TO_STAGE (immediately) as well.  All fine so far... it's like things "should" work.

  Unfortunately, upon return to the home page Manager ADDED_TO_STAGE, the graph becomes another ADDED_TO_STAGE parasite even though she is already in the stage's display list and was not REMOVED_FROM_STAGE in the meantime.  This sequence all occurs within the single function call to addChild (hand) in the root of the ADDED_TO_STAGE sprite Manager, it seems to be triggered by the Flash runtime behavior and not my code.

  Worse yet, I added managers to each of these three classes to look at ADDED and REMOVED as well as versions "to_stage" of the people, and I see a lot of redundant ADDED events with no DELETED at all events.

  Then a fundamental question: I'm wrong until only a single event ADDED_TO_STAGE for a given object, if it becomes an addChild() makes up for it and its parent and so on?

  Secondly, someone has a workaround or other ideas?  I had to make the installer in ADDED_TO_STAGE managers since I can't rely on the 'stage' of any DisplayObject property given as the value down here (other than the root sprite itself).

  At present, the only option seems to be moving out of ADDED_TO_STAGE and in constructors for everything except the root sprite.  Is that really the approach typical and/or required?

  At the present time, based on what I see the SDK 1.1.0 and running on OS 1.0.6 I would say that if everyone expects only one ADDED and ADDED_TO_STAGE event, and perhaps to build large parts of their user interface in these managers, they will want to verify if they actually build duplicates of many things and possibly raise a really messed up or sluggish UI accordingly.

  If necessary, I should be able to it will shrink down to a small sample that shows the problem.  I'll hold off doing this until I hear what others have to say.  Thank you!

  (Edit: tried to change the title of the topic.)

  Oh, just found this the problem report, noting this is an official bug as well.

  Several workaround solutions, each with different disabilities, including the removeEventListener() approach, are listed.

• No operational members and UCS vPC

  I get the following error message when you try to configure a vPC to a set of 6248 FIs to a set of N5Ks. Lacp and vpc are enabled on the 5ks.

  "No operational Member".

  Sho vpc br

  Legend:

  (*) - local vPC is down, transfer via vPC peer-link

  domain mail order ID: 1

  Status of peers: peers contiguity formed ok

  vPC KeepAlive status: peer is alive

  Consistency of configuration status: success

  State of coherence by vlan: success

  State of consistency of type 2: success

  role of the vPC: secondary

  Configured number of VPC: 1

  Peer gateway: disabled

  Duplicate assets excluded VLAN: -.

  Graceful consistency check: enabled

  The automatic recovery status: activated (timeout = 240 seconds)

  Sho vpc consistency-parameters int BP 22

  Legend:

  Type 1: vPC will be suspended in the event of incompatibility

  Name Type Local Peer value

  -------------               ----  ---------------------- -----------------------

  Do not close the Lan 1 no

  STP Port Type DEFAULT 1

  Port STP guard 1 none none

  STP STD simulate PVST 1 DEFAULT

  mode                        1     on                     on

  Speed 1 10 Gbit/s, 10 Gb/s

  1 full duplex

  Trunk to trunk for Mode 1 port

  VLAN 1 1 1 native

  MTU                         1     1500                   1500

  Port admin mode 1

  map of vPC type 1 empty empty

  Allowed VLAN - 1, xxx.xxx.xxx.xxx.xxx.xxx 1-4094

  Sho int BP 22 br

  --------------------------------------------------------------------------------

  VLAN port-channel Type Mode State reason speed Protocol

  Interface

  --------------------------------------------------------------------------------

  Trunk eth Po22 1 place no a - 10 G (D) no

  feat(Config-if) #.

  Sho run membership int BP 22

  ! Command: show running-config interface port-channel22 membership

  ! Time: Kills Sep 28 18:31:06 2009

  version 6.0 (2) N2 (2)

  interface port-channel22

  switchport mode trunk

  Speed 10000

  VPC 22

  interface Ethernet1/15

  switchport mode trunk

  channel-group 22

  interface Ethernet1/16

  switchport mode trunk

  channel-group 22

  Terry,

  I got the same error when starting to integrate UCS with the Nexus product line. In addition to LACP, how vPC that you have on "every" 5 K? In the above diagrams, you should have 2. I bet that you have only 1.

  Mike

• What is the difference between UCS - SD - 16 and UCS - SD - 16G =?

  Hello

  I have a few Cisco B200 m3 blades and will install 16 GB SD cards in them.  The documentation says that I need cards SD - UCS - 16G, but I received cards telling UCS - SD - 16 G =.  Does anyone know what, if any, difference the fact of '='?  They are the same card?  I have to make a large purchase of these cards and want to make sure that these are the correct type before ordering.  Thank you

  Hi @EBendixen,

  They are of the same type of cards are supported equal. The main difference is that:

  • SD cards with '=': are blank cards. The image that is required for the server can be downloaded on the map
  • SD cards without '=': are cards that come with the server image

  You can buy cards with this in mind. It all depends on if you already have the image of the server you need.

  Rgrds,

  Martin, computer scientist

• question about redundancy and failover from one site to tunnel

  First, I create the underside of crypto card,

  crypto IPSec_map 10 card matches the address encrypt-acl
  card crypto IPSec_map 10 set peer 209.165.201.1

  card crypto IPSec_map 10 the transform-set RIGHT value

  card crypto IPSec_map 10 the value reverse-road

  Then I set up a 2nd card statement, corresponding to the same ACL.

  crypto IPSec_map 20 card matches the address encrypt-acl
  card crypto IPSec_map 20 set peer 23.10.10.10

  card crypto IPSec_map 20 set transform-set RIGHT

  card crypto IPSec_map 20 set reverse-road

  My first question is - since Cryptography cards are processed in order, does that mean the first tunnel VPN (map 10) will always be used if its place?

  If so, what happens when the 209.165.201.1 remote peer becomes unreachable? is the tunnel of the 23.10.10.10 peer back automatically?

  What is the best way to reach a VPN site-to-site main and secondary where 209.165.201.1 primary school and 23.10.10.10 is backup and appears only when the primary is down?

  Thank you

  Hello

  As you mentioned that the Cryptography cards are processed in the order.

  If two cryptographic cards have the same "interesting" traffic then the second card encryption is never used (first crypto card is used).

  The best way to get a redundancy is to do the following:

  crypto IPSec_map 10 card matches the address encrypt-acl
  card crypto IPSec_map 10 set peer 209.165.201.1 23.10.10.10

  card crypto IPSec_map 10 the transform-set RIGHT value

  card crypto IPSec_map 10 the value reverse-road

  Note in the example above that you have defined an instance unique card crypto with two counterparts. The first pair will try first and if not answered, the second peer will be used as a backup.

  It will be useful.

  Federico.

• 6500 double and UCS

  Hi all

  We have recently installed a UCS enviornment with 2 x 6100 connected to our double heart 6509 devices. It is set up so that the 2 x 10 GigE rising connect of FAB - A the 6509-1 and 2 x 10 GigE connect b - FAB to 6509-2. Each of the 4 trunks environmental 6509 are configured similarly and barter same VLAN 4-FAB - A or B - FAB. We have 2 x UCS Chassis, each with 4 half-width servers. Each frame has a side A and B, with 20 GB goes to FAB - A and 20 GB will FAB - B.

  UCS environment above was installed by a 3rd party, while I set up the central part of the links.

  Our server team recently raised a Windows Server and I started light mac-beating on one of our hearts, one of the 10 GigE ports connected to FAB - A and link trunk between our 2 cores.

  I'm a little hesitant about how, on the side of 6509, the ports must be configured... "" Part 3 told me that there is no design "assets / standby", all trunks to each of the two 6100 s are active "all the time." Our 6509 do not run the VSS, just supervisors 'standard' sup720s running HSRP, with a trunk between the two.

  I understand there are no doubt a more details, but I said I would look into a perspective of networking/core and the server guy checks at its end, as in its present form it has the server in question related to 2 vNIC, one in each of the 6100, the two 'up '.

  Alan/Jason

  Looking at the wire, he speaks of a vswitch.

  For hypervisor type environments, (currently) it is recommended to have fabric OFF failover and Brad blog indicate that.

  Then turning on fabric failover will you all good and will have unpredictable failover of failure.

  A detailed topology (or a case of TAC) would help to hone in on the issue.

  A lot depends on how your vswitch is configured as before the typo traffic or other of the YEW, the vswitch must make a redirect which way out--the decision on A or B (give you 2 vNIC gave a vswitch which I think you would have). Once the traffic is delivered to a particular FI, and then it depends on how active uplink, it will take which could dictate whether the trunk between the cores get used etc.

  Hope that helps.

  Thank you

  -Matt

• Best way to move virtual machines between two clusters: HP and UCS

  Hello

  We move the virtual machines in the cluster UCSChassis: 6 guests ESXi in a Cluster of HPChassis.

  I wonder how do it with minimum downtime.

  Thanks for your help,

  If you have shared storage - expose the storahe where virtual machines are stored to both sets of guests - and the hosts are ocmpatible you can vmotion - or at least - to stop the mover VM's from the environment of the UCS and add it to the HP environment.

  If you can't share the storage you - you can use VMwate Standalone converter and V2V the VMs-

• ESXi 5.1 and UCS blades

  Hello

  I heard that there is a performance extremely questions degradedation with Vpshere5.1 currently running on the blades of the UCS, I wonder if anyone has confirm this or has any documented information on this. I am running vsphere5.0 and I really want to go to 5.1 but if there is really any questions I've heard that it can be in my best interest not to upgrade.

  Thank you!!!

  I have not heard this either.

• Normal redundancy and physical disks

  Version of the grid: 11.2.0.3
  Platform: Oracle Enterprise Linux 6
  
  
  I'm a bit confused about the term "Two-way mirrored" in the documentation of the DSO.
  
  
  If I want to keep 1 tera bytes of business data in an ASM diskgroup with NORMAL redundancy.

  Should I need 2 physical Disks of 1TB each ?
  or 
  Should I need 3 physical Disks of 1TB each ?

  ASM is not a software RAID. ASM works based on file extensions and provides data redundancy by using the free space on drive failure of disk groups. By default, each disc is a group of failure. For a mirror (normal) 2 channels, you need at least 2 rupture discs groups; for the 3 - way (top), you need at least 3. You can also create a 2-way mirror with 3 drives, in which case data will be written alternating. Having ASM normal redundancy in your example you must have at least two drives of 1 TB, but you can also use 3 and will have more space. You can also use an external RAID and use no redundancy ASM (external), although ASM will give you better performance and flexibility because you can add disks to increase the space without having to rebuild your storage space.