Knowledge Partner
Knowledge Partner
2759 views

Rolling Cluster Upgrade - removing old NetWare node?

According to the docs, you remove the old NetWare node from the tree, and then delete the Cluster Node, Server object and all corresponding objects related to the downed NetWare server.

I'm assuming by "cluster node" object they mean you go into the cluster object (the thingy that looks like 3 red balls) and find the "cluster" node that looks like an NCP server object with a red ball on it?

Secondly,

After you delete all those objects, are you still supposed to see it in the cluster?
Meaning you go into iManager, Cluster -> Cluster Manager, and it still shows that old server as belonging in the cluster, just that it's "unknown" state?
Labels (1)
0 Likes
6 Replies
warper2 Outstanding Contributor.
Outstanding Contributor.

Re: Rolling Cluster Upgrade - removing old NetWare node?

kjhurni wrote:

>
> According to the docs, you remove the old NetWare node from the tree,
> and then delete the Cluster Node, Server object and all corresponding
> objects related to the downed NetWare server.
>
> I'm assuming by "cluster node" object they mean you go into the cluster
> object (the thingy that looks like 3 red balls) and find the "cluster"
> node that looks like an NCP server object with a red ball on it?
>
> Secondly,
>
> After you delete all those objects, are you still supposed to see it in
> the cluster?
> Meaning you go into iManager, Cluster -> Cluster Manager, and it still
> shows that old server as belonging in the cluster, just that it's
> "unknown" state?
>
>


I have seen this happen before. The only way could fix it was bring the
whole cluster down and backup. Maybe someone has another way.


0 Likes
Knowledge Partner
Knowledge Partner

Re: Rolling Cluster Upgrade - removing old NetWare node?

Yeah, something royally screwed up

Cluster Options in iMangler shows the server and when you click on it, the cluster node shows up with an IP that's one less than what it really is.

So the new node won't insert/join into the cluster

Still waiting on NTS to look at my SR

I guess I'll give up for tonight and try again tomorrow or over the weekend if we have to take everything down. You'd think Novell would've gotten a better fix than having to down an entire cluster (like some sort of database fix without having to break everything).
0 Likes
Knowledge Partner
Knowledge Partner

Re: Rolling Cluster Upgrade - removing old NetWare node?

Well got an SR open at this point. Something definitely wrong with the cluster. eDir shows all the correct information, but when you go into iMangler -> Clusters -> Cluster Manager it has the following problems:

a) It shows a new resource called: CN (that has a triangle/pyramid icon) and shows "edir sync" with a yellow status icon. But you don't see this "resource" anywhere else.
b) The cluster node shows up, when you click on it, it shows it has an IP address that's one less than what it really is. (ie, 10.10.1.9 and it's really 10.10.1.10). It does this for EVERY cluster node

But, if you go into the Cluster Options, instead, and click on the cluster node, it shows the correct information there. eDir objects have the correct IP, etc.

I vaguely remember running into this years ago when we tried to remove/reinsert a cluster node on NetWare.

Where in the world does the cluster hold/get its information? It can't be from eDir because eDir shows up okay, so where the cluster thinks these other resources and IP's are is beyond me.
0 Likes
Knowledge Partner
Knowledge Partner

Re: Rolling Cluster Upgrade - removing old NetWare node?

Well, it's definitely a problem with either the docs or the cluster. I just removed a 2nd NetWare node as per the docs and it does the same thing. eDir doesn't show any traces of the old server object, yet the iManager -> Clusters -> Cluster Manager and Cluster Options still show the Cluster node object.

sigh.
0 Likes
Highlighted
Knowledge Partner
Knowledge Partner

Re: Rolling Cluster Upgrade - removing old NetWare node?

So it seems it's a combination of the Panning ID (which I fixed) and SP8. I guess the gipc.nlm in SP8 forces the NCS to read the Panning ID from eDir, whereas before it would read from a file or something.

Unfortunately the "Fix" is to adjust the Panning ID and then power everything down and bring it back up (grrrr)

So we'll see what happens Friday night when we take everything down.
0 Likes
Knowledge Partner
Knowledge Partner

Re: Rolling Cluster Upgrade - removing old NetWare node?

Well, that resolved the problem.

Unfortunately my OES2 node didn't want to load NCS automatically (or it did and errored out). I was able to get it to load manually
but then I couldn't migrate an NSS resource over to that node (it would try to load) but it would fail trying to add the secondary IP. And for some reason on OES2, it kept insisting on loading CIFS for the resource even though I never installed CIFS on the OES2 node and we never had it on NetWare.

I'll have to do more research but obviously the clustering docs aren't accurate (found a ton of errors and submitted feedback).
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.