On 07.12.2018 13:44, mathiasbraun wrote:
>
> Best bet, definitely. Now he has to decide which DIB the majority of
> current information holds. They've been divorced 14 weeks ago...
>
>
And likely administration has been done on both sides since then ;)

CU,
--
Massimo Rosen
Micro Focus Knowledge Partner
No emails please!
http://www.cfc-it.de

Just like concurrently administering groups in AD: one side wins.

On 07.12.2018 15:04, mathiasbraun wrote:
>
> Just like concurrently administering groups in AD: one side wins.
>
>
With AD, no side wins, it's all lost. ;)

--
Massimo Rosen
Micro Focus Knowledge Partner
No emails please!
http://www.cfc-it.de

Good point. But wait... MS wins: your money, your data, ...

On 07.12.2018 17:54, mathiasbraun wrote:
>
> Good point. But wait... MS wins: your money, your data, ...
>
>
There is that.. ;)

--
Massimo Rosen
Micro Focus Knowledge Partner
No emails please!
http://www.cfc-it.de

Hi Massimo,

Thanks for your input and suggestions. Looks like now both servers and hosting eDirectory. I was under the impression that eDirectory is only on primary server. I will try downing the servers and reconnecting them at a later time and update this thread with results. BTW, is there command to take backup of eDirectory data, delete eDirectory data, restore eDirectory from backup. If I could do that then I could just restore the data on secondary.

Every server in a tree hosts eDirectory. You can partition the tree and place replicas of these partitions on servers as you like, but even a server that doesn't hold any replica still hosts eDirectory. In your case i'd assume there is a total of two servers SUPPOSED to be in the tree and that there's only one partition (which is the [root] partition). For unknown reasons (but likely to those Massimo posted earlier in this thread) you ended up with a highly undesirable offset, in which LFA1 assumes to be the master of a tree with a second server called LFA2. While LFA2 considers itself to be the master of a single-server tree. Consequently, LFA1 tries to sync changes made in its database to LFA2, but as LFA2 doesn't know of LFA1 it ignores these sync attempts. On the other hand, if there are changes taking place in the DIB of LFA2, the latter one doesn't even try to sync them anywhere as it's not aware of any other server in its tree.
What this means in the first place is, that you have some sort of "split brain condidtion" ever since July 24th, i.e. if you created an object on LFA1 since then, LFA2 doesn't know anything about it. And, of course, vice versa.
The schematic plan (without details) to clean this up would be something like
- decide which server will survive in the first place
- backup data and trustee assignments from the other one
- try to export objects created on the other one after jul24th (e.g. via LDAP)
- wipe the other one (you can keep NSS volumes)
- clean up the tree on the survivor (i.e. clean up the replica ring and DIB from references to by now wiped box)
- patch the survivor to current code
- install the "other one" into the survivor's tree (PLEASE do so with current code, i.e. the stuff you've used one step before)
- try to recreate / import objects exported in step#3
- restore data / reattach data volumes
- restore trustee assignments as needed

Just a rough overview with no warranty. There are, of course, other ways to get this done, but the outline above can be followed without special tools.

Hi Mathias,

Could you explain how I do this ? - "clean up the tree on the survivor"
Should I open the objects browser in iManager and delete all objects that have LFA2 including LFA2. (could I use 3rd party LDAP tools to do this?)
After I clean up the tree , could I do this? Shutdown LFA2. Create another LFA2 server (with the same ip address as before ) and install it under the same tree as LFA1.
create volumes and restore data from backup tape.

I also tried some other method:
Because now we have bigger space available, I created a new server LFA1 and restored the eDirectory that I took from previous LFA1 server. ( I do not need another server LFA2. I have space for all the volumes of LFA2 in this LFA1). Then I removed all references to LFA2. Also recreated server and local certificates. I have all the users from the previous server, but i notice only one problem. If I create a volume, the volume is created but with an error message - NDS error 669. Also the volume does not show in Files/Folders in iManager client.

You would shutdown LFA2 first, then remove it (with ndsrepair) from the replica ring and finally wipe all objects referencing it from the DIB. Once finished you could install a new LFA2 into the tree, you could use the "old" IP again if you like.

As for the second part: 669 means "no access" which is likely a result of the way you "restored eDir from the previous LFA1 server". When you create a volume the system tries to create a corresponding volume object in NDS, there's a "servernameadmin" object (likely LFA1admin in your case), whose identity is used for this task. This object gets created on server installation and should have inheritable supervisory rights to the server's context. I'd assume that your restore procedure left you with a problem in this area. You'll likely be able to resolve this by re-running the configuration workflow for NSS via "yast2 nss".

Hi Mathias,

For the second part: I tried the yast2 nss (reconfigure). Even after this I am getting an error (-669) while creating a new Volume (volume is created except for the error message and it is not listed in iManager client). The ndstrace also says authentication failed. Anything else I can do to debug this.

Thanks for you help.

Srinivas.