Welcome Serena Central users! CLICK HERE
The migration of the Serena Central community is currently underway. Be sure to read THIS MESSAGE to get your new login set up to access your account.
jlmasmit Absent Member.
Absent Member.
405 views

Cluster installation


Hello,

We have a requirement to install Identity Manager 4.0.1 into a cluster.

Is there any supported installation of IDM4 in cluster? I haven't found
anything in the documentation and searching on the web I have found only
two references with old versions of IDM, SLES and Windows.

'Cool Solutions: AppNote: Clustering eDirectory and IDM on Windows
2003' (http://www.novell.com/coolsolutions/appnote/14856.html)

http://www.novell.com/support/php/search.do?cmd=displayKC&docType=kc&externalId=http--wwwnovellcom-communities-it-node-1537-configurailability20cluster20idm20320and20edirectory2088&sliceId=&docTypeID=DT_ARTICLES_TIPS_1_1&dialogID=308112846&stateId=0%200%20308114159

We need a supported installation with the current versions of Windows
(20008R2) and SLES (11) and with 64-bits operating systems.

Many thanks,
Regards,

Josep Lluís.


--
jlmasmit
------------------------------------------------------------------------
jlmasmit's Profile: http://forums.novell.com/member.php?userid=20336
View this thread: http://forums.novell.com/showthread.php?t=452837

Labels (1)
0 Likes
22 Replies
Knowledge Partner
Knowledge Partner

Re: Cluster installation

On Thu, 01 Mar 2012 17:36:01 +0000, jlmasmit wrote:

> We have a requirement to install Identity Manager 4.0.1 into a cluster.


Bug 748414 has been reported for the missing documentation.
(https://bugzilla.novell.com/show_bug.cgi?id=748414)

Not that that does you much more good than it does for me, I guess. I was
looking for the same thing.


> Is there any supported installation of IDM4 in cluster?


As far as I know, the docs for IDM36 still include the cluster install
directions, and should still work for IDM4. There's been no announcement
that I know of that clusters are no longer supported, so I assume that
they still are.


> We need a supported installation with the current versions of Windows
> (20008R2) and SLES (11) and with 64-bits operating systems.


I've not done a Windows cluster with eDir/IDM. I have done SLES clusters
(SLES10/Heartbeat), which works great. I'm looking in to how to upgrade
my 5 node SLES10/Heartbeat cluster to SLES11/HA.



--
--------------------------------------------------------------------------
David Gersic dgersic_@_niu.edu
Knowledge Partner http://forums.novell.com

Please post questions in the forums. No support provided via email.

0 Likes
TE Super Contributor.
Super Contributor.

Re: Cluster installation


dgersic;2178614 Wrote:
> On Thu, 01 Mar 2012 17:36:01 +0000, jlmasmit wrote:
>
> > We have a requirement to install Identity Manager 4.0.1 into a

> cluster.
>
> Bug 748414 has been reported for the missing documentation.
> (https://bugzilla.novell.com/show_bug.cgi?id=748414)
>
> Not that that does you much more good than it does for me, I guess. I
> was
> looking for the same thing.
>
>
> > Is there any supported installation of IDM4 in cluster?

>
> As far as I know, the docs for IDM36 still include the cluster install
> directions, and should still work for IDM4. There's been no
> announcement
> that I know of that clusters are no longer supported, so I assume that
> they still are.
>
>
> > We need a supported installation with the current versions of

> Windows
> > (20008R2) and SLES (11) and with 64-bits operating systems.

>
> I've not done a Windows cluster with eDir/IDM. I have done SLES
> clusters
> (SLES10/Heartbeat), which works great. I'm looking in to how to
> upgrade
> my 5 node SLES10/Heartbeat cluster to SLES11/HA.
>
>
>
> --
> --------------------------------------------------------------------------
> David Gersic
> dgersic_@_niu.edu
> Knowledge Partner
> http://forums.novell.com
>
> Please post questions in the forums. No support provided via
> email.


We are looking at doing IDM 4.0.1 in a SLES 11 SP1 HAE mode as well. We
raised the documentation issue with Novell Support. Here is the response
we received;

From an official positioning - IDM is cluster ready and is supported in
a cluster environment.

The documentation for “Installing Identity Manager in Clustering
Environment” was pulled from the IDM 3.6.1 documentation, because it
referenced Cool Solutions. Cool Solutions in and of themselves could
be written by Novell or third parties and are not supported by Novell
Technical Support.

IDM is dependent upon eDirectory. Thus documentation regarding
eDirectory's HA through clustering will need to be released prior IDM
documentation. The IDM documentation should talk about best practices
for deploying IDM in a cluster environment. Both these additions to
documentation are scheduled for eDirectory 8.8.7, and IDM 4.0.2
respectively.

So in a nutshell, IDM 4.0.1, is cluster ready, and the best source of
guides to implementing IDM clustering are the cool solutions as
referenced in the IDM 3.6.1 documentation, as there is no official
documentation or guide to implementing it at this time.

I read this as it is supported, but until eDir 8.8.7 and IDM 4.0.2, you
are on your own to figure out how to do it.

Here are my situation, so far.

We are doing this on VM's (VMWare Infrastructure). Server OS is SLES 11
SP1. eDir is 8.8.6, IDM is 4.0.1. Shared Storage will be SAN based,
don't know the SAN vendor.

We will be doing a 2 server cluster in Active/Passive mode for IDM with
Shared Storage.
'Configuring a Linux High Availability Cluster for IDM 3 and eDirectory
8.8 using SLES 10 and iSCSI'
(http://www.novell.com/communities/it/node/1537/configuring%20linux%20high%20availability%20cluster%20idm%203%20and%20edirectory%208.8)

We will be doing a 2 server cluster in Active/Active mode for the LDAP
Tree only clustering the eDir2eDir Driver, which is one-way from IDM
only.
'IDM Clustering on SLES Without Shared Storage'
(http://www.novell.com/communities/node/8386/idm-clustering-sles-without-shared-storage)

Anybody have any idea how to do STONITH in this environment? Not sure
where the servers reside in the VMWare Farm, they may be on any host,
and I don't know how many hosts there are in DEV or PROD. We need to be
able to handle any number of VMWare Hosts.

Any other concerns? Differences between SLES 10 Heartbeat and SLES 11
HAE that are going to cause issues?

Management is curious if eDir could be clustered in Active/Active mode
using Shared Storage. Not sure how you could do this using two separate
Servers using one copy of eDir, but maybe someone has figured out how
(David?).


--
tse7147
------------------------------------------------------------------------
tse7147's Profile: http://forums.novell.com/member.php?userid=4730
View this thread: http://forums.novell.com/showthread.php?t=452837

0 Likes
Anonymous_User Absent Member.
Absent Member.

Re: Cluster installation


To all,
Here is a document with links to other articles foe each of your
needs.
'Important Notice'
(http://www.novell.com/communities/node/6937/novell-identity-manager-high-availability-concepts)

Cheers,
Leroy Joseph
Visual Click Software
(eDirectory Management and Reporting)
'eDirectory Management | DSRAZOR for eDirectory'
(http://www.visualclick.com/content/dsrazor-for-edirectory.htm)


--
leroyjjr
------------------------------------------------------------------------
leroyjjr's Profile: http://forums.novell.com/member.php?userid=75462
View this thread: http://forums.novell.com/showthread.php?t=452837

0 Likes
Knowledge Partner
Knowledge Partner

Re: Cluster installation

On 3/16/2012 1:26 PM, leroyjjr wrote:
>
> To all,
> Here is a document with links to other articles foe each of your
> needs.
> 'Important Notice'
> (http://www.novell.com/communities/node/6937/novell-identity-manager-high-availability-concepts)


Gee, I had totally forgotten about that article. Wonder if it needs
updating for IDM4?

Hmm...


0 Likes
TE Super Contributor.
Super Contributor.

Re: Cluster installation


geoffc;2182901 Wrote:
> On 3/16/2012 1:26 PM, leroyjjr wrote:
> >
> > To all,
> > Here is a document with links to other articles foe each of your
> > needs.
> > 'Important Notice'
> > ('Important Notice'

> (http://www.novell.com/communities/node/6937/novell-identity-manager-high-availability-concepts))
>
> Gee, I had totally forgotten about that article. Wonder if it needs
> updating for IDM4?
>
> Hmm...


What, you don't have enough other articles in work to keep you busy?

Why is it, whenever a link to a Cool Solutions article is referenced
here, it shows up as "Important Notice" instead of the article title?
Why is it, if the Roundabouts in England go clockwise, does the sun
still rise in the east there?


--
tse7147
------------------------------------------------------------------------
tse7147's Profile: http://forums.novell.com/member.php?userid=4730
View this thread: http://forums.novell.com/showthread.php?t=452837

0 Likes
Knowledge Partner
Knowledge Partner

Re: Cluster installation

On Fri, 16 Mar 2012 13:36:02 +0000, tse7147 wrote:

> We are looking at doing IDM 4.0.1 in a SLES 11 SP1 HAE mode as well. We
> raised the documentation issue with Novell Support. Here is the response
> we received;
>
> From an official positioning - IDM is cluster ready and is supported in
> a cluster environment.


That looks really familiar. I think I read essentially the same message
last week.


> We are doing this on VM's (VMWare Infrastructure). Server OS is SLES 11
> SP1. eDir is 8.8.6, IDM is 4.0.1. Shared Storage will be SAN based,
> don't know the SAN vendor.


I have a physical cluster with SAN, and a VM based cluster with iSCSI.
These are SLES10/HA. Both work fine. I'm working on upgrading to SLES11/
HA.

The first problem you need to solve is access to the storage. I don't
know how your SAN and VM host are going to interact here.


> We will be doing a 2 server cluster in Active/Passive mode for IDM with
> Shared Storage.
> 'Configuring a Linux High Availability Cluster for IDM 3 and eDirectory
> 8.8 using SLES 10 and iSCSI'


Yeah, basically that works fine. No worries there. Except I'd go with
three nodes so you have a quorum.


> We will be doing a 2 server cluster in Active/Active mode for the LDAP
> Tree only clustering the eDir2eDir Driver, which is one-way from IDM
> only.


Also fine, but again, I'd go with three nodes for a quorum.


> Anybody have any idea how to do STONITH in this environment?


You'll need to investigate what's possible. The Linux HA project has a
STONITH agent for VMWare, if I recall correctly. In SLES11/HA, there's
also the SBD (Split Brain Detection) support, which may help you as well.


> Any other concerns? Differences between SLES 10 Heartbeat and SLES 11
> HAE that are going to cause issues?


The major changes are all improvements. Most of the Heartbeat2 (SLES10)
stuff still applies, even if the implementations have changed somewhat.


> Management is curious if eDir could be clustered in Active/Active mode
> using Shared Storage. Not sure how you could do this using two separate
> Servers using one copy of eDir, but maybe someone has figured out how
> (David?).


Absolutely not. You can put multiple eDirectory instances up in the
cluster, but each one gets their own DIB.



--
--------------------------------------------------------------------------
David Gersic dgersic_@_niu.edu
Knowledge Partner http://forums.novell.com

Please post questions in the forums. No support provided via email.

0 Likes
TE Super Contributor.
Super Contributor.

Re: Cluster installation


dgersic;2182918 Wrote:
> On Fri, 16 Mar 2012 13:36:02 +0000, tse7147 wrote:
>
> > We are looking at doing IDM 4.0.1 in a SLES 11 SP1 HAE mode as well.

> We
> > raised the documentation issue with Novell Support. Here is the

> response
> > we received;
> >
> > From an official positioning - IDM is cluster ready and is supported

> in
> > a cluster environment.

>
> That looks really familiar. I think I read essentially the same
> message
> last week.
>
>
> > We are doing this on VM's (VMWare Infrastructure). Server OS is SLES

> 11
> > SP1. eDir is 8.8.6, IDM is 4.0.1. Shared Storage will be SAN based,
> > don't know the SAN vendor.

>
> I have a physical cluster with SAN, and a VM based cluster with iSCSI.
> These are SLES10/HA. Both work fine. I'm working on upgrading to
> SLES11/
> HA.
>
> The first problem you need to solve is access to the storage. I don't
> know how your SAN and VM host are going to interact here.



It seems to be working however the VMWare guy built it. It looks like
it is just a vmdk that is visible to both nodes but mounted on neither.
I am using an ocf/filesystem resource to mount/dismount it and that
does appear on whatever node I migrate the resource to.

>
>
> > We will be doing a 2 server cluster in Active/Passive mode for IDM

> with
> > Shared Storage.
> > 'Configuring a Linux High Availability Cluster for IDM 3 and

> eDirectory
> > 8.8 using SLES 10 and iSCSI'

>
> Yeah, basically that works fine. No worries there. Except I'd go with
> three nodes so you have a quorum.


We were given two nodes for each eDir instance (IDM and LDAP), and just
asking for another one for the PostgreSQL server for Reporting caused
too many "ripples" in the space/time continuum...
>
>
> > We will be doing a 2 server cluster in Active/Active mode for the

> LDAP
> > Tree only clustering the eDir2eDir Driver, which is one-way from IDM
> > only.

>
> Also fine, but again, I'd go with three nodes for a quorum.
>
>
> > Anybody have any idea how to do STONITH in this environment?

>
> You'll need to investigate what's possible. The Linux HA project has a
> STONITH agent for VMWare, if I recall correctly. In SLES11/HA, there's
> also the SBD (Split Brain Detection) support, which may help you as
> well.


We upset the space/time continuum to ask for a small partition for SBD,
and that actually happened after much consternation. So, that is what we
will be using. One 1 gig partition for both clusters.

>
>
> > Any other concerns? Differences between SLES 10 Heartbeat and SLES

> 11
> > HAE that are going to cause issues?

>
> The major changes are all improvements. Most of the Heartbeat2
> (SLES10)
> stuff still applies, even if the implementations have changed
> somewhat.


I like what I have seen so far.

>
>
> > Management is curious if eDir could be clustered in Active/Active

> mode
> > using Shared Storage. Not sure how you could do this using two

> separate
> > Servers using one copy of eDir, but maybe someone has figured out

> how
> > (David?).

>
> Absolutely not. You can put multiple eDirectory instances up in the
> cluster, but each one gets their own DIB.


Well, I found this interesting resource model;

'ocf_heartbeat_eDir88'
(http://linux-ha.org/doc/man-pages/re-ra-eDir88.html)

Not sure where it came from, but I would really like to know where
they plan on going with this. Also would like to find out more about how
it works.

I thought this line in the document was amusing...

-If heartbeat times out before eDir has had a chance to start properly,
mayhem *WILL ENSUE*.-


So far I have built the STONITH/SBD, VIP (ocf_heartbeat_ipaddr2) and
Shared Storage (ocf_heartbeat_filesystem) resources. STONITH seems to be
throwing some kind of errors, not sure where those are logged. The VIP
and Shared Storage seem to be working. At least the IP Address is
active, and df -h shows the storage moving.

Now to dig into that eDir88 resource...

More to follow.


--
tse7147
------------------------------------------------------------------------
tse7147's Profile: http://forums.novell.com/member.php?userid=4730
View this thread: http://forums.novell.com/showthread.php?t=452837

0 Likes
Knowledge Partner
Knowledge Partner

Re: Cluster installation

On Thu, 22 Mar 2012 16:56:02 +0000, tse7147 wrote:

> It seems to be working however the VMWare guy built it. It looks like
> it is just a vmdk that is visible to both nodes but mounted on neither.


Ok, I haven't worked with that, specifically, but as long as you can use
the ocf/filesystem resource agent to mount and unmount it, you should be
fine. Depending on the file system you're using, you may want to use
'tune2fs' to prevent it from periodically being fsck'd on mount, with, of
course, the usual caveats applied to disabling that particular feature.

>> Yeah, basically that works fine. No worries there. Except I'd go with
>> three nodes so you have a quorum.

>
> We were given two nodes for each eDir instance (IDM and LDAP)


So you have four, total? Or two? Either way, odd numbers are better, and
it's best if you can establish a quorum (>2 nodes).


> asking for another one for the PostgreSQL server for Reporting caused
> too many "ripples" in the space/time continuum...


Ah. Politics. Can't help you there.


>> You'll need to investigate what's possible. The Linux HA project has a
>> STONITH agent for VMWare, if I recall correctly. In SLES11/HA, there's
>> also the SBD (Split Brain Detection) support, which may help you as
>> well.

>
> We upset the space/time continuum to ask for a small partition for SBD,
> and that actually happened after much consternation. So, that is what we
> will be using. One 1 gig partition for both clusters.


Ah, ok, cool. I haven't done the SLES11/HA with SBD support yet.


>> > Any other concerns? Differences between SLES 10 Heartbeat and SLES

>> 11
>> > HAE that are going to cause issues?

>>
>> The major changes are all improvements. Most of the Heartbeat2 (SLES10)
>> stuff still applies, even if the implementations have changed
>> somewhat.

>
> I like what I have seen so far.

Yeah, it works really well here. There's a learning curve, like with
anything else, but once you get past that and have the first couple of
resource groups running, it gets a lot easier.


>> > Management is curious if eDir could be clustered in Active/Active

>> mode
>> > using Shared Storage. Not sure how you could do this using two

>> separate
>> > Servers using one copy of eDir, but maybe someone has figured out

>> how
>> > (David?).

>>
>> Absolutely not. You can put multiple eDirectory instances up in the
>> cluster, but each one gets their own DIB.

>
> Well, I found this interesting resource model;
>
> 'ocf_heartbeat_eDir88'
> (http://linux-ha.org/doc/man-pages/re-ra-eDir88.html)

Yes, that's the one you want for managing eDir instances, works great.


> Not sure where it came from, but I would really like to know where they
> plan on going with this. Also would like to find out more about how it
> works.


Like any OCF, it's job is to start the resource, monitor the resource,
and stop the resource.

Start is the equivalent of 'ndsmanage start...' As long as you have
everything else needed in place, the instance starts, and off you go.

Stop is also the equivalent of 'ndsmanage stop...', and again, there's
not usually much drama here.

Monitor is the interesting case. Is the instance up, running, and
working? If not, HA tries to fix it, by restarting it or moving it to
another node.


> I thought this line in the document was amusing...
>
> -If heartbeat times out before eDir has had a chance to start properly,
> mayhem *WILL ENSUE*.-


If the resource misbehaves, the node it's attempting to start on could be
fenced. It then attempts to move to another node, which could then be
fenced. And around the cluster you go. If you don't have good constraints
set up, it could kill all nodes. That counts as mayhem.


--
--------------------------------------------------------------------------
David Gersic dgersic_@_niu.edu
Knowledge Partner http://forums.novell.com

Please post questions in the forums. No support provided via email.

0 Likes
TE Super Contributor.
Super Contributor.

Re: Cluster installation


dgersic;2184505 Wrote:
> On Thu, 22 Mar 2012 16:56:02 +0000, tse7147 wrote:
>
> > It seems to be working however the VMWare guy built it. It looks

> like
> > it is just a vmdk that is visible to both nodes but mounted on

> neither.
>
> Ok, I haven't worked with that, specifically, but as long as you can
> use
> the ocf/filesystem resource agent to mount and unmount it, you should
> be
> fine. Depending on the file system you're using, you may want to use
> 'tune2fs' to prevent it from periodically being fsck'd on mount, with,
> of
> course, the usual caveats applied to disabling that particular
> feature.
>
> >> Yeah, basically that works fine. No worries there. Except I'd go

> with
> >> three nodes so you have a quorum.

> >
> > We were given two nodes for each eDir instance (IDM and LDAP)

>
> So you have four, total? Or two? Either way, odd numbers are better,
> and
> it's best if you can establish a quorum (>2 nodes).
>
>
> > asking for another one for the PostgreSQL server for Reporting

> caused
> > too many "ripples" in the space/time continuum...

>
> Ah. Politics. Can't help you there.
>
>
> >> You'll need to investigate what's possible. The Linux HA project has

> a
> >> STONITH agent for VMWare, if I recall correctly. In SLES11/HA,

> there's
> >> also the SBD (Split Brain Detection) support, which may help you as
> >> well.

> >
> > We upset the space/time continuum to ask for a small partition for

> SBD,
> > and that actually happened after much consternation. So, that is what

> we
> > will be using. One 1 gig partition for both clusters.

>
> Ah, ok, cool. I haven't done the SLES11/HA with SBD support yet.
>
>
> >> > Any other concerns? Differences between SLES 10 Heartbeat and

> SLES
> >> 11
> >> > HAE that are going to cause issues?
> >>
> >> The major changes are all improvements. Most of the Heartbeat2

> (SLES10)
> >> stuff still applies, even if the implementations have changed
> >> somewhat.

> >
> > I like what I have seen so far.

>
> Yeah, it works really well here. There's a learning curve, like with
> anything else, but once you get past that and have the first couple of
> resource groups running, it gets a lot easier.
>
>
> >> > Management is curious if eDir could be clustered in Active/Active
> >> mode
> >> > using Shared Storage. Not sure how you could do this using two
> >> separate
> >> > Servers using one copy of eDir, but maybe someone has figured out
> >> how
> >> > (David?).
> >>
> >> Absolutely not. You can put multiple eDirectory instances up in the
> >> cluster, but each one gets their own DIB.

> >
> > Well, I found this interesting resource model;
> >
> > 'ocf_heartbeat_eDir88'
> > ('ocf_heartbeat_eDir88'

> (http://linux-ha.org/doc/man-pages/re-ra-eDir88.html))
>
> Yes, that's the one you want for managing eDir instances, works great.
>
>
> > Not sure where it came from, but I would really like to know where

> they
> > plan on going with this. Also would like to find out more about how

> it
> > works.

>
> Like any OCF, it's job is to start the resource, monitor the resource,
> and stop the resource.
>
> Start is the equivalent of 'ndsmanage start...' As long as you have
> everything else needed in place, the instance starts, and off you go.
>
> Stop is also the equivalent of 'ndsmanage stop...', and again, there's
> not usually much drama here.
>
> Monitor is the interesting case. Is the instance up, running, and
> working? If not, HA tries to fix it, by restarting it or moving it to
> another node.
>
>
> > I thought this line in the document was amusing...
> >
> > -If heartbeat times out before eDir has had a chance to start

> properly,
> > mayhem *WILL ENSUE*.-

>
> If the resource misbehaves, the node it's attempting to start on could
> be
> fenced. It then attempts to move to another node, which could then be
> fenced. And around the cluster you go. If you don't have good
> constraints
> set up, it could kill all nodes. That counts as mayhem.
>
>
> --
> --------------------------------------------------------------------------
> David Gersic
> dgersic_@_niu.edu
> Knowledge Partner
> http://forums.novell.com
>
> Please post questions in the forums. No support provided via
> email.

I am working two problems right now. The first is errors that show up
on the monitoring for the STONITH resource when I do crm status. Here is
what I see;

============
Last updated: Thu Mar 22 17:20:59 2012
Stack: openais
Current DC: nim2t7 - partition with quorum
Version: 1.1.2-2e096a41a5f9e184a1c1537c82c6da1093698eb5
2 Nodes configured, 2 expected votes
4 Resources configured.
============

Online: [ nim1t7 nim2t7 ]

StoneMe (stonith:external/sbd): Started nim2t7
VIP (ocf::heartbeat:IPaddr2): Started nim2t7
Storage (ocf::heartbeat:Filesystem): Started nim2t7

Failed actions:
StoneMe_monitor_0 (node=nim2t7, call=133, rc=1, status=complete):
unknown error
eDir_start_0 (node=nim2t7, call=156, rc=-2, status=Timed Out):
unknown exec error
eDir_start_0 (node=nim1t7, call=23, rc=-2, status=Timed Out):
unknown exec error

All the resources, that are working correctly, and can be verified via
the console. But the STONITH monitor is not happy. As you can see, the
eDir resource is not happy at all, it is not working. eDir is working
properly on the nodes, when I do ndsmanage startall/stopall, if the VIP
and Storage are on that server, it works. That is good on either node.
I just cannot get it to work via the resource (which is the second
problem).

Still climbing the learning curve on this. Need to know where the logs
are so I can pick them apart.


--
tse7147
------------------------------------------------------------------------
tse7147's Profile: http://forums.novell.com/member.php?userid=4730
View this thread: http://forums.novell.com/showthread.php?t=452837

0 Likes
TE Super Contributor.
Super Contributor.

Re: Cluster installation


I think I have the eDirectory problem sorted out. We'll see, still
testing.

What I did was setup some Cluster Constraints.
First, I set the preferred node for all the resources to node 1.
Then I set the co-location to the VIP, Storage and eDir resource such
that they all must be running on the same node
Finally, I setup a load order so that VIP loads first, then Storage and
finally eDir. Sorta hard to load eDir when the DIB is not available.

Still working out all of the Constraint settings, there are a few that
need more explanation than what is in the manager app. But, if I
deactivate nodes, things appear to be moving.

Next step is to install the IDM package. That should not affect the
cluster as that loads and unloads with eDir. There are two settings in
the eDir cluster resource that are interesting. Both have to do with
assuring services are started, for LDAP and IDM. I have IDM set false
currently (doh, no IDM installed!!!).


--
tse7147
------------------------------------------------------------------------
tse7147's Profile: http://forums.novell.com/member.php?userid=4730
View this thread: http://forums.novell.com/showthread.php?t=452837

0 Likes
Knowledge Partner
Knowledge Partner

Re: Cluster installation

On Fri, 23 Mar 2012 13:06:01 +0000, tse7147 wrote:

> What I did was setup some Cluster Constraints. First, I set the
> preferred node for all the resources to node 1. Then I set the
> co-location to the VIP, Storage and eDir resource such that they all
> must be running on the same node Finally, I setup a load order so that
> VIP loads first, then Storage and finally eDir. Sorta hard to load eDir
> when the DIB is not available.


Mostly right. The constraint for node preference is good, as it tells you
where things will "normally" be running. You don't have to, but I like to
do it that way. But instead of colocation constraints, make a Group. The
group then contains your VIP, Disk, and eDir resources. That way they all
act together, and in the correct order.


> Next step is to install the IDM package. That should not affect the
> cluster as that loads and unloads with eDir.


You'll want to do this carefully... When the IDM installer stops the eDir
instances, the cluster monitor notices, and restarts them for you. The
best way I've found to do this is to take the cluster down, start the
instances manually, then install IDM.


--
--------------------------------------------------------------------------
David Gersic dgersic_@_niu.edu
Knowledge Partner http://forums.novell.com

Please post questions in the forums. No support provided via email.

0 Likes
TE Super Contributor.
Super Contributor.

Re: Cluster installation


Another SNAFU... 😞

The eDir88 resource was working fine, until I installed IDM. Now
things have really gone south. One node went totally unclean, and won't
come back. The eDir88 resource seems to only want to start on the
unclean node, and cannot because the other resources are on the working
node.

TGIF!!!


--
tse7147
------------------------------------------------------------------------
tse7147's Profile: http://forums.novell.com/member.php?userid=4730
View this thread: http://forums.novell.com/showthread.php?t=452837

0 Likes
Knowledge Partner
Knowledge Partner

Re: Cluster installation

On Fri, 23 Mar 2012 17:46:01 +0000, tse7147 wrote:

> Another SNAFU... 😞
>
> The eDir88 resource was working fine, until I installed IDM. Now things
> have really gone south. One node went totally unclean, and won't come
> back.


You have to install IDM on both nodes. See the -DCLUSTER_INSTALL=TRUE
command line switch.


> The eDir88 resource seems to only want to start on the unclean
> node, and cannot because the other resources are on the working node.


You need a group, then the cluster will move the entire group, not just
the offending resource.


--
--------------------------------------------------------------------------
David Gersic dgersic_@_niu.edu
Knowledge Partner http://forums.novell.com

Please post questions in the forums. No support provided via email.

0 Likes
TE Super Contributor.
Super Contributor.

Re: Cluster installation


dgersic;2184817 Wrote:
> On Fri, 23 Mar 2012 17:46:01 +0000, tse7147 wrote:
>
> > Another SNAFU... 😞
> >
> > The eDir88 resource was working fine, until I installed IDM. Now

> things
> > have really gone south. One node went totally unclean, and won't

> come
> > back.

>
> You have to install IDM on both nodes. See the -DCLUSTER_INSTALL=TRUE
> command line switch.
>
>
> > The eDir88 resource seems to only want to start on the unclean
> > node, and cannot because the other resources are on the working

> node.
>
> You need a group, then the cluster will move the entire group, not
> just
> the offending resource.
>
>
> --
> --------------------------------------------------------------------------
> David Gersic
> dgersic_@_niu.edu
> Knowledge Partner
> http://forums.novell.com
>
> Please post questions in the forums. No support provided via
> email.


I did the IDM install on both nodes as indicated, one direct, one
cluster. For those who have not done the cluster install, it looks the
same as a regular install, it just does not ask for an admin ID.

Moved it all to a group, the VIP, Storage and eDir. Still no luck
making it work with the cluster. VIP and Storage move back and forth
with ease, but the eDir88 resource won't load. I am able to go to the
node that has the VIP and Storage and load eDir from the command line
(ndsmanage startall) without any problems. Some configuration item is
missing. My first thought is that the eDir88 resource is starting to
quickly, before the address and storage have been brought up, but even
after they have been running a while, the cluster still will not start
it.


--
tse7147
------------------------------------------------------------------------
tse7147's Profile: http://forums.novell.com/member.php?userid=4730
View this thread: http://forums.novell.com/showthread.php?t=452837

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.