SRM failover scenarios

Similar Questions

• IPCC Enterprise - Server Cti failover: scenario

  Hi, I have a duplex scales environment (loggerA, routerA, PG1A, CG1A and CTIOSServerA)

  B-side: (loggerB, routerB, PG1B, CG1B and CTIOSServerB)

  AW, Distributor and HDS are on a stand-alone server.

  I tried the script to fail in the next and both CTIServer (CG1A and CG1B) did not CTIOSServer (AorB) to connect (based on the scenario of Services running).

  My failover of the scenario:

  -On the scales I m running services: LoggerA, RouterA, PG1A, and CTIOSA

  -On the b-side: I m running only the services of CG1B

  All Ports or IP are well done I m wondering if the failover scenario is supported, because until the PG1B has not elapsed the CG1B does not CTIOSA you want to connect.

  Reading of the CSCma27627 case (symptom: no documentation published, noting a specific configuration)

  requirements at the CTI server on dedicated platform installation/deployment

  independent of the same instance of PG.

  Condition: PG and CTI Server is installed on a separate server platforms

  Workaround: No. This configuration is not permitted.)

  So do you mean CG1B is not able to connect to a PG1A?

  Concerning

  The system is designed such that one side of the CG A and B aren't only for the PG on his own side. The current process of mutual FUNDS on the PG will not support 2 servers CTI pointing to it to any side. CTI servers can be independent active on the side of the PIM is enabled when the two CSPs are online.

  So in your example, you must add mutual FUND and PIM next B in order to make it able to communcate with PIM on the a-side.

• SRM Reprotect scenarios

  Hi, ive done some research on SRM that we could implement in a few months.  Had questions about the part reportect of things and some scenarios.

  With the release of SRM 5.1, it now allows reprotect when you use replication of Vsphere.  But from what ive read it only allows it if he were a planned migration not a disaster.  Not sure if this is true, if anyone knows?

  So I was wondering what is happening in some scenarios.  All scenarios are with replication of Vsphere

  
  Lets say the power outage on the main site.  We discover from the power company something bad has happened and power might be down for more than 24 hours.  They decide to use the DR site and switch to it.  Everything works very well after a bit of work, people are working from home or another location.

  After 24 hours power returns to the other site.  We want the failback to the primary site.  So what is happening here.  If reprotect works then just restore us, but if reprotect only works in planned migration then we can just push the button to reprotect.

  We must therefore create a recovery plan, return to the main site.  This part, I don't have a problem with until you get to the part where you want to use the vmdk files that are already on the main site.  But this raises questions

  With Vsphere replication, there may be a delay of 15 minutes between the sites.  Which means the Main Sites (vmdk) virtual machine might have 15 minutes of data that has not been synchronized on the site of DR.  I don't like to get these data in only 15 minutes, but when I goto to reseed these vmdk files to different to use?

  Basically im trying to avoid having to resynchronize the vmdk together via the wan connection, or else copy the VMDK on the site of Dr. and dissemination on say via USB drive and copy them to the main site then.  It would take to long (some disasters require it of course)

  Is smart enough RS to just trash the last 15 minutes that does not get synced to the top and start to synchronize new data on the DR site?

  If that's enough, smart whats a typical cut off the coast before its not able to do it, I can see situations with bad outages of the internet where users can continue to work for an hour or two before decide us actually tipping, in these cases is that long?

  I hope it made sense

  Thank you

  Mike

  You can run 'Restore' operation after a recovery plan was complete. It does not matter that the stimulus plan was executed in planned Migration or recovery mode after a disaster. Usually planned migration is performed when the production and dr site are rising and running and there are some predicted the failure, while the disaster recovery mode is used in case of unexpected events.

  When you configure the VM to be replicated with vSphere Repliacation, one of the parameters that you specify is the RPO (recovery point objective), it can very from 15 minutes to 24 hours. It's time that is acceptable for your company to have a loss of data. In your example, when you click Reprotect, SRM will reverse replication in the opposite direction and production of old VMS files vmdk will be the initial seed. This will cause the changes that have taken place on the site of DR since the event of failover to replicate in return. VR will conduct a parity check and will copy only the differences that occurred in activity on the DR site to the primary site of origin. You don't need to manually download files in the site Dr. back to the original site. Of couse, the more changes on these files on the site of Dr., data more than you need to synchronize again, but, again, it will be only the changed data, not the entire file.

• SRM failover 5 but no back

  I recently had an interesting application that I never really thought.

  Is it possible to use SRM for failover, but not automatic and manual restoration cleaning DR site?

  I have never thought about it because it's quite unusual that in theory, if the data does not change, why not have a session watching Dr. but the scenario that gave me was that (thanks, hurrican Sandy):

  Prod site loses MPLS, but the servers are still operational.

  They want to put in place an instance of the application (data) on the DR site for other remote sites to at least be able to access the data, while the primary site can continue to operate in this way.

  When MPLS is in place, they can reorient their production sites and continue as usual and do a manual cleaning of SRM on the site of DR.

  I saw not all the documentation on the issue of whether this is possible or not.  I hope to keep protection associations and as intact as possible stimulus plans without having to cross and recreate it.  Someone at - it experience with this or can direct me to a blog or documentation that can get me going in the right direction?

  Thanks in advance!

  Jeff

  No problem.

  This is what communities are for.

• Site Recovery Manager failover scenario

  Hi guys,.

  I'm going to the planning of a Cluster of ESX installation two different place in the same building.

  I'm trying to archive a RTO less than 8 minutes, my solution is based on two site with vcenter its own SRM server and array based replication.

  I have a simple question, if I have to test a recovery plan, I can go to the source site and start the recovery for a single virtual machine or group plan, but what happens if my site has completely fails? In the one site I have all the VM Virtual e Center the SRM Server Reed.

  The procedure starts automatically from an entry of site "at a distance"? Or should I enter in the remote site and manually start a remote plan?

  Thank you

  Note that disaster recovery is not equal to high availability and to declare a disaster that many things must take into account and people from other regions must be involved. Anyway, here you can find a way to automate failover: Automation of failover with SRM - VMware Articles

• SRM failover failed with RecoveryPoint

  Hi people,

  We have table RecoveryPoint SRM installation. We have not had any problems with any configuration. But when we have validated the failover we get the following error at recovery steps.

  -Impossible to synchronize the data on the replica consistency group ' *'. Replication is not active for some reason any in the copy of the group.

  I have attached the report of the history of recovery plan for reference. Maybe someone knows what could cause this problem?

  FOR INFO. In RP consistency group strategy, we put "SRM" and managed by "External Application" (this is what our team of storage done if I had no idea what it means). Also I don't know if this is what the origin of the problem.

  Any help would be really appreciated.

  Thank you and best regards,

  Hari.

  We opened the case with VMware and finally settled the issue. Issue was not with RP or RS. It comes with the services of ESXi. When trying to download anything from the data in the local client store it throws the error 'Operation timed out' immediately once start us the download. Then we tried restart ESXi services services.sh restart.for guests. Then everything started working so far. This isn't the same problem reported in known issues in MRS.

  Hope this helps someone.

  Thank you and best regards,

  Hari.

• Why the return of flame is necessary in the failover scenario?

  Hello. Why must he get on flashback on main databases and backup for failover to work?

  The primary must be repaired to restore.  Flashback is a way to do it. DataGuard scenarios

  The day before he must after becoming a primary and then failing back to become a standby.

• SRM failover for virtual machines but not all VMDK disks to include

  Hi, I have a number of VMs that need failover protection, but not all the VMDK disks must be included when switching to the opposite site. For example, a virtual machine can have 3 discs and players only need 1 & 2, switch to the recovery site - SRM can be configured to ignore the 3rd disc?

  Thank you

  I was not aware of a fact that replied vmdk cannot be detached. I think that this is normal.

  Why you are replicating data, you do not need? You can place these VMDK on LUN not replicated?

• SRM failover real without bringing down production

  Hi all, I'm under ESX 3.5 U5 and SRM 1.0 Update1. I did a few tests SRM and they all go well. We now have a real failover on our site of DR. The DR site is completely separate from production. The only difference in this real shift is that we will use it as a test, but we do not want RS to stop the virtual machines in the protected area; My question is: can we the power down the protected site of Virtual Center, so there is no communication between the protected site of VC and VC site recovery and avoid powering down of the protected virtual machines? What we have done and has worked in the past is a firewall on the environment of production completely and the VC of recovery site is therefore not able to communicate to the protected site VC and eventually do not power off the coast of the VM protected in production.

  What we're trying to accomplish is to have our failure on VMs in DR communicate with other servers physical and material on the site of DR (completely isolated from production). Unless we can test RS, but instead of creating the network of bubbles recovered virtual computers have access to the DR test so they can access all devices in the vLAN Dr.

  Any input will be greatly appreciated. I always read about how to configure the vSwitch to connect Dr. vLAN and see if that meets our needs.

  Thank you

  Hello

  I'm sure that if protected vCenter is interrupted during failover, VMs will not stop, cause SRM server do not communicate directly, all communication goes through vCenter servers. The stage of 'Shutdown VMs' expires.

  I do not understand completely the second part of your post (on the connection to DR VLAN). Why you want to achieve and in which situation (test / real failover)?

  Michael.

• How granular is SRM failover?

  I know that this is probably a simple question - I've stuck with marketing jargon and just am not not clear on that.

  SRM seeks to recover a site or he can do and also failover individual VMs?

  By example, if I deploy:

  Site A - 40 mV, 4 physical servers, SAN, data stores 4

  Site B - 40 mV, 4 physical servers, SAN, data stores 4

  If I lose 1 server in site a can I served it to Site B or should it be all 40 virtual machines, or entire data store?

  Y at - it a failback option?

  Finally, is it versioning/snapshots, as well as replication (which is managed by the SAN I know not, but are available VM snapshots)?

  Thank you!

  There are some interesting points here...

  First of all, the others in this thead element are correct - SRM failovers all virtual machines in the data store - really the only way to recovery on the VM data store that contains 20 are to have a fair recovery plan for her - place you 19 of the virtual machines in the area "not lit" and the unique virtual machine in the normal priority. The get together 20 recovered during the test - but effectively only a VM would be active. It is a clumsy work-around that could not serve your purpose - and useless if you press run...

  We will have to wait until the berries are VM-ware - where the storage array 'knows' that the volumes/LUNS are formatted for VMFS/NFS and detect the vmx/vmdk file type. Things are going in this direction - the new Clarrion CX4 and Navisphere give you excellent visablity right up to the size and the type of virtual disk – Navisphere VM...

  With regard to what constitutes a failure. To some people will say that the loss of a data store could not be considered as disaster - given that can be recovered by a local kind of snapshot to the table, rather than replication to another geographic location. As ever I guess depends on the size and scope of your infrastructure...

  Finally, automatic failover. With the SDK and some work with .net, it is possible to automate the recovery plans. But I'm his terriritory dangerous - because you could could easily get splitting-brain/false positive. If you look at this kind of availability - maybe a HA cluster stretched with something like Metro of NetApp cluster is was really, should be considered as...

  Concerning

  Mike Landry

  RTFM education

  http://www.RTFM-ed.co.UK

  Author of the book of MRS:http://www.rtfm-ed.co.uk/2010/03/22/new-administrating-vmware-site-recovery-manager-4-0/

  Free PDF or printed at price

• Failover scenario when you have more than one standby DB

  DB version: 11.2.0.4

  OS: Linux/Unix

  I've only worked with Data Guard inverter set when there is only a standby DB.

  Consider the following configuration:

  Primary DB: orcl

  Physical standby DB1: orcl_sby0

  Physical standby DB2: orcl_sby1

  I understand that, if you want to do failover of SQL * Plus, the following command must be run on the primary database

  SQL > ALTER DATABASE COMMIT to SWITCH STANDBY mode PHYSICAL;

  What Eve DB will become the principal now?

  This command does not switch to digital. It is simply your primary database to start to behave in physical waiting.

  After that, you can choose either Standby (standby1 or standby2) 2 to pass the primary role by running "alter database commit to switching to the primary with the end of the session.

  So it is to you to choose which standby database you want to start to behave like a primary database and then set the log_archive_dest_ FAL and settings accordingly.

  -Jonathan Rolland

• Licensing vCenter with SRM

  Hi all

  I seem to be going round in circles with this question - I have read the documentation and recorded a call with the service the customer on this subject, and the answers are contradictory to say the least...

  We seek to implement a SRM solution in a new infrastructure - essentially a management cluster will use SRM failover management of virtual machines to a domain protected to a DC recovery in another city controller.  The DC recovery will be dark for normal operations, only more direct when MRS. failover occurs and will not host virtual machines.

  The question I have vCenter licenses of MRS. The documentation states that, during failover of a vCenter server from a site protected for a recovery site hosting any other virtual machines, only 1 vCenter license is required. However the VMware licenses team said the 2 licenses are required for automatic restoration of VMs for the protected site of origin once he recovered.

  So, who is? Any thoughts anyone (who is not a dealer or someone trying to sell me licenses)? Happy qualify points if necessary to answer the question...

  See you soon.

  Jeremy.

  the vcenter in Dominican Republic must be put under tension and linked with the SRM server of the RD, in this scenario, it's actually light and 2nd license is necessary.

• Mappings related inventory system SRM question

  Hello

  I was curious to know what the expected behavior should be with the inventory of the network related mappings in the following example scenario:

  Group mappings of protection previously configured at the protected Site

  Recovery plan already set up on recovery Site

  I am currently connected to the virtual Center on the "Protected" site server and access the mappings of the inventory of my groups that I have configured previously. I now want to change the virtual network vswitches that uses these virtual machines protected in a failover scenario in respect of the Dr. I created a new virtual switch on the site of DR already before entering the SRM tab on the protected site.

  At this point, I have to do to change the vswitch that will be used on all vms preserved my at a newly created the DR site. The change is successful without error.

  I now have to connect to the virtual Center Server at my place of DR and go to one place holder vm objects which is part of the recovery plan and note that it is still configured to use the previous vswitch just change once connected on the protected site.

  My question is this. These should automatically place holder VMs has been updated to use the new vswitch configuration? If they were to be updated automatically, what would be the reason why they are still configured to use the old vswitch?

  If they were not to be updated automatically, is still once it is reasonable to assume that I'll just have to "Edit configuration" on each of the VM reserved space to use the new vswitch and everything should be good?

  Thanks for any info you can provide

  Hello

  Existing placeholders are not supposed to get automatically updated when they change the mappings of an inventory. Mappings of the inventory are 'default' for new protected virtual machines.

  If I remember correctly, recovered VMs will be attached to the 'old' portgroup during a failover.

  I believe that the same is true when you manually reconfigure the placeholder.

  Probably, you need to restore the virtual machines.

  Michael.

• standby ip addresses? are required on all interfaces monitored for failover

  Hi all

  I need clarification on an interesting question that I observed during the configuration of an active installation / standby to be able to use 2 x 5525 cisco with version 8.6;

  Here is the configuration, we have 4 subnets that we need to keep separate. I have each of the ASAs connected to different subnets. However, only 1 subnet's IP address configured standby while all other subnets have only an active address on the active firewall. As this is a failover scenario, I have 2 interfaces for LAN and stateful failover.

  I just test the failover on 2 subnets without any standby ip address and to my surprise, everything seems to work as expected. Just need for clarification on why we need sleep on the monitored interfaces addresses when clearly the installer can work without any configured. Are there implications with instance without standby ip addresses?

  Thank you

  Especially at your facility can happen many things cannot be recognized by the ASA without a correct installation of failover. This could be a port of mafunctioning in your infrastructure for example.

  But leave approach it the other way around: what advantages do you see in the implementation in a non-standard way? Or what kind of problems do you expect? Usually the night before IP is not configured if there is no IP address for example on the outside interface.

  --
  Don't stop once you have upgraded your network! Improve the world by lending money to low-income workers:
  http://www.Kiva.org/invitedBy/karsteni

• ASA 5540 Stateful failover routing errors

  Hello

  Having two 5540's configuration in a failover scenario. Make the LAN failover and failover state. * See attachment *.

  Failover LAN use 192.168.2.1 as active and 192.168.2.2 as before, with the subnet mask of 30. On both LAN failover use G0/2 and there is a crossover cable connecting them.

  The failover of the State uses 192.168.3.1 as active and 192.168.3.2 as before, with the subnet mask of 30. With "enable HTTP replication" checked in ASDM. On both devices State failover uses G0/3 and there is a crossover cable connecting them.

  The ASDM syslog connects errors every 10 seconds or so to say that:

  SOURCE IP ADDRESS: 192.168.3.1

  DESTINATION IP: 192.168.3.2

  Description:

  "Routing could not locate the next hop for igrp NP identity 192.168.3.1/0 in statefull:192.168.3.2/0".

  The ASA use static routes to meet the network, these roads, there are two, and both are in the 10.x.x.x network. No routing protocol is in use.

  I don't know why these errors are "spamming" my syslog and would like to get rid of them.

  Glad to hear that it works, that's the most important thing. I don't mean to preach, but Cisco does not recommend using ADJUSTABLE wires to fail on. Devices cannot always say that the captain should be and usually causes questions more than a simple link to the bottom.