ONTAP Hardware

Strech metrocluster upgrade to 10 Gbps

IMPI
3,176 Views

Hi all,

 

i have a strech metrocluster composed like that :

 - 2 FAS 8020

 - 2 DS2246 (24x900) mirrored

 - 2 DS2246 (15x900) not mirrored 

 

As this is a "strech", there is no ATTO, only many fibers between them.

 

It is connected using e0e and e0f (1 Gbps)

Two SVM on each controller : first is using e0e as home port, e0f as failover, second is the opposite

 

We'd like to go 10 Gbps and then using e0a and e0b.

 

First of all : can i plugged the Gbic while FAS running like in switch ? We will use twinax cable (in fact gbic are integrated in it)

 

Then how to migrate : add e0a and e0b into failover groups, then set e0a as home for first SVM and e0b for second then do a "network interface revert -vserver * -lif *" to have lif going to new home port ?

 

 

In fact here is what i have, i tried to understand before writing but want to be sure :

 

 

Some info about the setup, do not read all this is boring 😊 i put in red all what is important

We are talking about controler A, this will be symetrical on B

 

 

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

 

clusterA::> network interface show
Logical Status Network Current Current Is
Vserver Interface Admin/Oper Address/Mask Node Port Home
----------- ---------- ---------- ------------------ ------------- ------- ----
SVM-A-01
SVM-A-01_nfs_lif1
up/up 192.168.251.211/24 clusterA-01 e0e true
SVM-A-02
SVM-A-02_nfs_lif1
up/up 192.168.251.212/24 clusterA-01 e0f true
SVM-B-01-mc
SVM-B-01_nfs_lif1
up/down 192.168.251.221/24 clusterA-01 e0e true
SVM-B-02-mc
SVM-B-02_nfs_lif1
up/down 192.168.251.222/24 clusterA-01 e0f true
clusterA
clusterA-01-ic1
up/up 192.168.224.221/24 clusterA-01 e0e true
clusterA-01-ic2
up/up 192.168.224.223/24 clusterA-01 e0f true
clusterA-01_mgmt1
up/up 192.168.240.71/24 clusterA-01 e0M true
cluster_mgmt up/up 192.168.240.61/24 clusterA-01 e0M true
8 entries were displayed.

 

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

 

clusterA::> network interface show -failover
Logical Home Failover Failover
Vserver Interface Node:Port Policy Group
-------- --------------- --------------------- --------------- ---------------
SVM-A-01
SVM-A-01_nfs_lif1 clusterA-01:e0e system-defined dataA
Failover Targets: clusterA-01:e0e, clusterA-01:e0f
SVM-A-02
SVM-A-02_nfs_lif1 clusterA-01:e0f system-defined dataA
Failover Targets: clusterA-01:e0f, clusterA-01:e0e
SVM-B-01-mc
SVM-B-01_nfs_lif1 clusterA-01:e0e system-defined Default
Failover Targets: clusterA-01:e0e, clusterA-01:e0f,
clusterA-01:e0a, clusterA-01:e0b
SVM-B-02-mc
SVM-B-02_nfs_lif1 clusterA-01:e0f system-defined Default
Failover Targets: clusterA-01:e0f, clusterA-01:e0a,
clusterA-01:e0b, clusterA-01:e0e
clusterA
clusterA-01-ic1 clusterA-01:e0e local-only interclusterA
Failover Targets: clusterA-01:e0e, clusterA-01:e0f
clusterA-01-ic2 clusterA-01:e0f local-only interclusterA
Failover Targets: clusterA-01:e0f, clusterA-01:e0e
clusterA-01_mgmt1 clusterA-01:e0M local-only managementA
Failover Targets: clusterA-01:e0M
cluster_mgmt clusterA-01:e0M local-only managementA
Failover Targets: clusterA-01:e0M

 

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

 

clusterA::> network interface failover-groups show
Failover
Vserver Group Targets
---------------- ---------------- --------------------------------------------
clusterA
Default
clusterA-01:e0M, clusterA-01:e0a,
clusterA-01:e0b, clusterA-01:e0e,
clusterA-01:e0f
dataA
clusterA-01:e0e, clusterA-01:e0f
interclusterA
clusterA-01:e0e, clusterA-01:e0f
managementA
clusterA-01:e0M
4 entries were displayed.

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

As you can see i have failover group dataA and interclusterA

   -> should i add e0a and e0b in both ? Not sure intercluster need big throughput... i thing this is only management, data are going through dedicated fiber

 

Is the command ok : network interface failover-groups add-targets -failover-group dataA -vserver SVM-A-01 -targets clusterA-01:e0a,clusterA-01:e0b

Same for SVM-A-02 (and symetrical to do on the other controler)

Then i should change home port, is that the right command :  network interface modify -vserver SVM-A-01 -lif SVM-A-01_nfs_lif1 -home-port e0a (and svm-A-02 with port e0b)

 

And then if i should  do that for intercluster :

With this command  ?  network interface modify -vserver clusterA -lif clusterA-01-ic1 -home-port e0A (and also ic2 with e0b)

 

Is all of that doable while running ? I guess yes as i only add more ports, then migrate home... The question is more important about can i plug the cable while running !

 

And then, at least, i guess i should remove e0e and e0f  from failover group where there is now e0a and e0b

 

Hope this was not to complicated to read, i tried to work the best by myself to only have a few questions 

 

thanks a lot

 

 

 

1 ACCEPTED SOLUTION

aladd
3,114 Views

Hello @IMPI 

 

No, there is no problem with connecting cables while the system is running.

 

I would recommend validating from the node that it registers the new connection before continuing with the configuration.

 

After the configuration has been completed, it would be recommended to remove the old ports from the failover groups, unless you wish for them to remain as a failover target.

View solution in original post

4 REPLIES 4

aladd
3,115 Views

Hello @IMPI 

 

No, there is no problem with connecting cables while the system is running.

 

I would recommend validating from the node that it registers the new connection before continuing with the configuration.

 

After the configuration has been completed, it would be recommended to remove the old ports from the failover groups, unless you wish for them to remain as a failover target.

IMPI
3,068 Views

Hi Aladd,

 

many thanks for this reply, this is great.

 

just only one thing i did not understand : "I would recommend validating from the node that it registers the new connection before continuing with the configuration."

 

Have a nice day

IMPI
2,997 Views

Hi Aladd,

 

could you specify what you meant ?

 

Thanks

IMPI
2,763 Views

Hi, for those who'd like to do this.

 

We did the upgrade and all went ok.

 

2 thinks to notice :

       on my command, cluster name was not good, it should be clusterA, not SVM... we can see it when doing  a "network interface failover-groups show" 

      i had to migrate lif but also the one with "-mc" (metrocluster)

 

and then i removed the 1Gbe interfaces leaving intercluster failover group unchanged

 

HTH !

 

Public