topic New Cluster Failover in Server Clustering

New Cluster Failover

GhiranoHPE — Tue, 14 Sep 2021 15:57:53 GMT

for the first time we manage to deploy a cluster failover using :

2 DL380 GEN10 servers

1 MSA 2062 10gb iscsi sff

2 hpe switch 1950 12xgt 4sfp+

we actually don't understand how to connect cables from servers to switches and from server to SAN

also we would understand how to set the two switches in fault tolerance

thank you

Re: New Cluster Failover

Ihaqueit — Tue, 14 Sep 2021 14:52:53 GMT

For failover cluster, Checkout this video.

(391) MicroNugget: What is Failover Clustering in Windows Server 2012? - YouTube

If you running any application or any services and meantime one server or nodes unavailable, cluster failover provides another server as as a main data server.

2 hpe switch 1950 12xgt 4sfp+.

In failover cluster once you see may be its running on separate ip of your nodes. When you setup the failover mechanism best way to provide separate vlan or ip, one for server and another for user.

Its a most adaptive technology.

You must be learn it about failover mechanism.

Re: New Cluster Failover

GhiranoHPE — Tue, 14 Sep 2021 15:03:49 GMT

we will use vmware essential plus to manage the cluster

just we would be sure to connect correctly the hardware, mostly i wonder how to connect lan interfaces from DL380 to two switches without losing connection

Re: New Cluster Failover

Ihaqueit — Tue, 14 Sep 2021 15:09:35 GMT

how to connect lan interfaces from DL380 to two switches without losing connection.

I am little bit confuse.

1. Is one dl380 you want to connect to two switches?

Re: New Cluster Failover

GhiranoHPE — Tue, 14 Sep 2021 15:24:54 GMT

not, we have two dl380 each ahs a 10gb either card with 2 lans SFP

so two 10 gb lan SFP for each server

MSA 2062 has 4 SFP port for each controller, total 8 ports

and two switches 1950 with 4 SFP each

Re: New Cluster Failover

Ivan_B — Tue, 14 Sep 2021 15:27:11 GMT

Hi @GhiranoHPE !

From the Networking point of view here are my suggestions:

1. Merge 2 x 1950 into single IRF stack.
2. Update the stack to the latest version
3. Enable STP just in case
4. VERY IMPORTANT STEP: Make all ports that are connected to servers 'STP edge' ports. Otherwise you will have traffic interruptions 30 seconds after each transition to UP state.

The rest is highly dependent on your servers configuration and how your servers will handle multiple link - BAGG (link aggregation with LACP) or switch-independent teaming. But from the point of view of hardware fault tolerance keep each server connected to BOTH switches, so if one switch fails, another will continue forwarding traffic.

Re: New Cluster Failover

sbhat09 — Tue, 14 Sep 2021 15:29:36 GMT

Hello @GhiranoHPE,

Please find the schematic diagram of the connections for your requirement/resources.

For more details you may find these documents useful:

https://support.hpe.com/hpesc/public/docDisplay?docId=a00105314en_us&docLocale=en_US

https://support.hpe.com/hpesc/public/docDisplay?docId=a00017710en_us

Regards,
Srinivas Bhat

If you feel this was helpful please click the KUDOS! thumb below!
Note: All of my comments are my own and are not any official representation of HPE.

Re: New Cluster Failover

GhiranoHPE — Wed, 15 Sep 2021 06:14:33 GMT

wow,

i would ask, what about the four gigabit lan for each server ? how do i connect them between the two switches ?