Anonymous_User Absent Member.
Absent Member.
1763 views

Adding a node (VM, RDM, OES2018) to an existing custer (OES2015SP)

Hello,

I am trying to add a node to an existing cluster:

existing: 2 nodes OES 2015 SP1, old Blades
new: ESXi 6.5 VMs, OES 2018, volumes mapped as RDM disks

Disks are visible on the new server, nssmu shows correct NSS pools, sbdutil
-f finds the SBD partition.

But when I start NCS the server immediately crashes; this results in a
reboot-loop until I unmap the SBD volume from the VM.

My goal is to move the cluster resources to 2 new nodes and then remove the
old ones. Do I miss something? Why does the node crash as soon as it sees
the SBD partition?

Thanks,
Mirko
Labels (1)
0 Likes
11 Replies
Anonymous_User Absent Member.
Absent Member.

Re: Adding a node (VM, RDM, OES2018) to an existing custer(OES2015SP)

Hello,

are all of the cluster nodes in the same subnet of the LAN?
Did you setup the new cluster according to "Novell Cluster Services
Implementation Guide for VMware" (clus_vmware_lx.pdf)?

Regards Burkhard
0 Likes
Anonymous_User Absent Member.
Absent Member.

Antw: Re: Adding a node (VM, RDM, OES2018) to an existingcuster(OES2015SP)


Hello,

yes, all nodes in the same Subnet.

I followed clus_vmware_lx.pdf as far as possible; the document is quite old,
the VMware GUI is quite different in 6.5, there are more options. And I did
not set up RDM disks at the command line. But I think effectively I did the
same things they are describing in the document.

One major difference is: I don't have exclusive NICs for my OES cluster
nodes.

Regards,
Mirko


>>>

> Hello,
>
> are all of the cluster nodes in the same subnet of the LAN?
> Did you setup the new cluster according to "Novell Cluster Services
> Implementation Guide for VMware" (clus_vmware_lx.pdf)?
>
> Regards Burkhard

0 Likes
Anonymous_User Absent Member.
Absent Member.

Re: Antw: Re: Adding a node (VM, RDM, OES2018) to an existingcuster(OES2015SP)

Hello,

there is TID7005837 on how to collect debug information of a cluster. If
you have the opportunity to ask microfocus support please do so. We
planned to migrate to virtual cluster nodes but at last we did not.

Regards Burkhard
0 Likes
Anonymous_User Absent Member.
Absent Member.

Re: Antw: Re: Adding a node (VM, RDM, OES2018) to anexistingcuster(OES2015SP)

Hello Burkhard,

thanks for the TID, wasn't aware of that.

May I ask what your considerations are for not migrating to VMs?

Thanks,
Mirko


>>>

> Hello,
>
> there is TID7005837 on how to collect debug information of a cluster. If


>
> you have the opportunity to ask microfocus support please do so. We
> planned to migrate to virtual cluster nodes but at last we did not.
>
> Regards Burkhard

0 Likes
aferris Absent Member.
Absent Member.

Re: Antw: Re: Adding a node (VM, RDM, OES2018) to anexisting

This sounds like my issue in this very (sub) forum. VMWare, OES 2018, NCS, etc. As soon as my node tries to touch the SBD disk it panics. I have OES 2011 and 2015 clusters so this isn't my first go at things.

Andrew
0 Likes
Anonymous_User Absent Member.
Absent Member.

Re: Antw: Re: Adding a node (VM, RDM, OES2018) to anexistingcuster(OES2015SP)

Hello,

we did not move to virtual clusters because it is more difficult to find
errors in case storage/VMWAre performance issues or hardware errors (Yes
the docu of novell requires "dedicated" NICs for the cluster nodes). We
have asked a novell supporter who cares about ncs for 20 customers. Only
one of them has virtual cluster nodes according to the novell
documentation. It works but it may be more difficult to maintain.

Regards Burkhard
0 Likes
Anonymous_User Absent Member.
Absent Member.

Antw: Adding a node (VM, RDM, OES2018) to an existing custer(OES2015SP)


There was a defect in NCS, I got a patch from Micro Focus and it works now.

Regards,
Mirko



>>>

> Hello,
>
> I am trying to add a node to an existing cluster:
>
> existing: 2 nodes OES 2015 SP1, old Blades
> new: ESXi 6.5 VMs, OES 2018, volumes mapped as RDM disks
>
> Disks are visible on the new server, nssmu shows correct NSS pools,
> sbdutil
> ‑f finds the SBD partition.
>
> But when I start NCS the server immediately crashes; this results in a
> reboot‑loop until I unmap the SBD volume from the VM.
>
> My goal is to move the cluster resources to 2 new nodes and then remove
> the
> old ones. Do I miss something? Why does the node crash as soon as it
> sees
> the SBD partition?
>
> Thanks,
> Mirko

0 Likes
dbgallo1 Absent Member.
Absent Member.

Re: Antw: Adding a node (VM, RDM, OES2018) to an existing cu

Great!

Is the patch in the upgrade channel or only available via support?

Tips....

We've been running NCS in virtual on vmaware for over 6 years, using RDM's ....

Our new servers ( CISCO UCS chassis) only have 10G nics so we didn't dedicate any to the NCS cluster...

Our setups are typical , don't forget the to read up on the grub menu changes that need to be made in 2015sp1 and earlier. ...

Each virtual server is locked to one host via affinity rules, no VMotion , no SRM . Do NOT run more than one server in a ncs cluster on one host
0 Likes
Anonymous_User Absent Member.
Absent Member.

Antw: Re: Adding a node (VM, RDM, OES2018) to an existing custer(OES2015SP)


I think, currently the patch is only available via support. It is expected
to be in "Update 2 for OES 2018".

If you open a SR you might want to ask for
ncs-kmp-default-2.6.5_k4.4.74_92.38-30.3.1395.5.PTF.1080760.x86_64.rpm

Regards,
Mirko

>>>


> Great!
>
> Is the patch in the upgrade channel or only available via support?
>
> Tips....
>
> We've been running NCS in virtual on vmaware for over 6 years, using
> RDM's ....
>
> Our new servers ( CISCO UCS chassis) only have 10G nics so we didn't
> dedicate any to the NCS cluster...
>
> Our setups are typical , don't forget the to read up on the grub menu
> changes that need to be made in 2015sp1 and earlier. ...
>
> Each virtual server is locked to one host via affinity rules, no VMotion
> , no SRM . Do NOT run more than one server in a ncs cluster on one host


0 Likes
Anonymous_User Absent Member.
Absent Member.

Antw: Re: Adding a node (VM, RDM, OES2018) to an existing custer(OES2015SP)



>>>


> Do NOT run more than one server in a ncs cluster on one host



I have read this somewhere else - but I did not understand it.Sure, most
probably it does not make much sense to run more VMs in one NCS cluster on
the same host - but if you emphasize the "NOT" there is probably a more
serious reason, one that I am not aware of yet?

Thanks,
Mirko
0 Likes
aferris Absent Member.
Absent Member.

Re: Antw: Re: Adding a node (VM, RDM, OES2018) to an existin

I am happy to hear there's going to be a patch.

I don't understand this line though:

"Do NOT run more than one server in a ncs cluster on one host" - Is it due to SRM, as I have a 2011 and a 2015 cluster with cluster nodes on separate ESXi hosts and the virtual servers/ccluster volumes move between them no problem. I agree that I can't vMotion the cluster nodes but that's okay. I also have done test clusters all on one ESXi host to let me use VMDKs and that is supported but unwise. 🙂

Andrew
0 Likes
The opinions expressed above are the personal opinions of the authors, not of Micro Focus. By using this site, you accept the Terms of Use and Rules of Participation. Certain versions of content ("Material") accessible here may contain branding from Hewlett-Packard Company (now HP Inc.) and Hewlett Packard Enterprise Company. As of September 1, 2017, the Material is now offered by Micro Focus, a separately owned and operated company. Any reference to the HP and Hewlett Packard Enterprise/HPE marks is historical in nature, and the HP and Hewlett Packard Enterprise/HPE marks are the property of their respective owners.