I've got a 2 node active/passive cluster and would like to replace one node with a new server. I've evicted the node to be replaced - what do I do next? Specifically how is the shared storage handled when adding that new server to ther cluster. Thanks in advance. Rick
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations derfloh on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
MSCS Replace Node
- Thread starter dumbkopf
- Start date
Is nothing special. Indeed you have to evict the node, then you can uninstall cluster service rom that node (evicted).
Shut down the alive node.
Then you bring a new computer, you will make the physical connections as was with the previous node (SCSI bus, connect it to the shared device). You have to follow same rules, like: take care that the SCSI id's must be different on that bus, terminator issues, etc.
Then you startup the first node, and the second node too. In that moment they will not share any drive. You have to install cluster service on the second node, and then, during install to specify that you want a "join an existing cluster".
Again, during install you need to have the right IP addresses for your node (the private network connection and the public one).
As about applications, then it depends on the applications, and in which model the cluster was implemented:active-active, active-pasive. So, for a[pplication better study for each of them how the uninstall of a node is working.
Gia Betiu
gia@almondeyes.net
Computer Eng. CNE 4, CNE 5, MCSE Win2K
Shut down the alive node.
Then you bring a new computer, you will make the physical connections as was with the previous node (SCSI bus, connect it to the shared device). You have to follow same rules, like: take care that the SCSI id's must be different on that bus, terminator issues, etc.
Then you startup the first node, and the second node too. In that moment they will not share any drive. You have to install cluster service on the second node, and then, during install to specify that you want a "join an existing cluster".
Again, during install you need to have the right IP addresses for your node (the private network connection and the public one).
As about applications, then it depends on the applications, and in which model the cluster was implemented:active-active, active-pasive. So, for a[pplication better study for each of them how the uninstall of a node is working.
Gia Betiu
gia@almondeyes.net
Computer Eng. CNE 4, CNE 5, MCSE Win2K
- Thread starter
- #3
Gia,
In regards to your response you wrote 'Shut down the alive node.' Is there any way around this (i.e. in a real world production scenario shutting down the 'alive node' cuts off user access to the surviving cluster member)? If at all possible I would like to avoid/minimize impact to the end-users while a cluster node is being replaced. Thanks.
In regards to your response you wrote 'Shut down the alive node.' Is there any way around this (i.e. in a real world production scenario shutting down the 'alive node' cuts off user access to the surviving cluster member)? If at all possible I would like to avoid/minimize impact to the end-users while a cluster node is being replaced. Thanks.
Sorry, i was in holiday.
Why to shutdown the active node? The reason: SCSI connections. But, if your SCSI supports hot plug then shouldn't be a problem. I want to mention that is not a condition of the Cluster Service or OS. Is just because you have to make some chages on the SCSI bus.
Gia Betiu
gia@almondeyes.net
Computer Eng. CNE 4, CNE 5, MCSE Win2K
Why to shutdown the active node? The reason: SCSI connections. But, if your SCSI supports hot plug then shouldn't be a problem. I want to mention that is not a condition of the Cluster Service or OS. Is just because you have to make some chages on the SCSI bus.
Gia Betiu
gia@almondeyes.net
Computer Eng. CNE 4, CNE 5, MCSE Win2K
I also am in the process of replacing both nodes of an active passive cluster.
I'll be doing the following, so as to have as little impact to users as possible.
I am however using HBA attached Dell Array 650F via a switch, so i have no direct SCSI Bus attached disks and can see the array system from the new server.
If your using direct attached (non hot plug) disks i would recommend shutting down the cluster and seeing if you can access the disks from your replacement node. if you can access the disks then the steps below will work for you.
Build first new server
Uninstall Cluster service from secondary node
Evict current secondary node (standby node) from primary node using cluster manager
Add new server to cluster
Fail over to new node.
Fail back to old node
Fail over to new node
Evict old node (should be in standby)
Add second new node
Fail over and back.
DONE.
I'll be doing the following, so as to have as little impact to users as possible.
I am however using HBA attached Dell Array 650F via a switch, so i have no direct SCSI Bus attached disks and can see the array system from the new server.
If your using direct attached (non hot plug) disks i would recommend shutting down the cluster and seeing if you can access the disks from your replacement node. if you can access the disks then the steps below will work for you.
Build first new server
Uninstall Cluster service from secondary node
Evict current secondary node (standby node) from primary node using cluster manager
Add new server to cluster
Fail over to new node.
Fail back to old node
Fail over to new node
Evict old node (should be in standby)
Add second new node
Fail over and back.
DONE.
ITsmyfault
IS-IT--Management
Hi -
I'm in a very similar boat. 2 node Win2k cluster. C1 got corrupt somehow, administrative user did not have rights to patch it etc.. wierd stuff. So we rebuilt it. Evicted old C1 from cluster, built new C1. Installed cluster services and told it to join the existing cluster. We got various errors doing that so I shut everything down after hours. When I brought it back up, the new C1 thinks it's in a cluster but C2 (the former "good" host) cannot start cluster services bc it says it cannot authenticate. I checked and double checked the cluster user password and it's fine.
Any ideas? I'm staring at re-installing MSCS on C2 now. I just don't to keep going back and forth!
I'm in a very similar boat. 2 node Win2k cluster. C1 got corrupt somehow, administrative user did not have rights to patch it etc.. wierd stuff. So we rebuilt it. Evicted old C1 from cluster, built new C1. Installed cluster services and told it to join the existing cluster. We got various errors doing that so I shut everything down after hours. When I brought it back up, the new C1 thinks it's in a cluster but C2 (the former "good" host) cannot start cluster services bc it says it cannot authenticate. I checked and double checked the cluster user password and it's fine.
Any ideas? I'm staring at re-installing MSCS on C2 now. I just don't to keep going back and forth!
You changed the cluster service account password. Bring both nodes down at the same time, then bring up node 1, then node 2.
next time follow
next time follow
- Status
Similar threads
- Locked
- Question
- Locked
- Question