[Freeipa-users] Different Database Generation ID

[Ludwig Krispenz]

Hi,

you get the "different database generation" if one side is built from 
scratch or reimported from a plain ldif without repl stat e information. 
replication will only work if both sides have the same data origin.

About initlializing back and forth it depends on your topology if it can 
become a problem. If a replica is reinitialized it's changelog is 
recreated (the old one will no longer match) and if you do it again in 
the other direction you remove the changelog there as well - and then 
can msis changes not yet replicated to other replicas and you can run 
into the "csn not found problems".

I looked up one of your previous posts about which version of 389-ds you 
are using, and it looks like you have one we know has some issues, as 
stated several times on this list :-(

About your observation that replication is stopping and working again 
after restarting, this can be a problem of the replication agreement 
going into fatal state instead of retrying. Restarting the server 
overcomes this, but you could achieve it by disabling the agreement.

Ludwig

On 10/11/2016 06:13 PM, Ian Harding wrote:
> I have this error in the log of my FreeIPA server freeipa-sea.bpt.rocks:
>
> [11/Oct/2016:09:04:39 -0700] NSMMReplicationPlugin -
> agmt="cn=masterAgreement1-seattlenfs.bpt.rocks-pki-tomcat"
> (seattlenfs:389): The remote replica has a different database generation
> ID than the local database.  You may have to reinitialize the remote
> replica, or the local replica.
>
> So I did this:
>
> ipa-replica-manage re-initialize --from freeipa-sea.bpt.rocks
>
> on seattlenfs
>
> But the error continues.
>
> I think I know why.  freeipa-sea had a meltdown and I had to rebuild it,
> and established it as a replica of seattlenfs.  Unfortunately, I think
> seattlenfs was a replica of the original freeipa-sea.
>
> It seems like a bad idea to reinitialize themselves from each other, and
> in fact it's warned against here:
>
> https://access.redhat.com/documentation/en-US/Red_Hat_Directory_Server/8.2/html/Administration_Guide/Managing_Replication-Troubleshooting_Replication_Related_Problems.html
>
> "... Also, M2 should not initialize M1 back."
>
> But in looking at my bash history I have indeed done that as well.
>
> Is there any way out of this mess?  These two servers actually DO
> replicate, most of the time.  They stop for no reason and restarting the
> ipa services on freeipa-sea does get them started again.
>

-- 
Red Hat GmbH, http://www.de.redhat.com/, Registered seat: Grasbrunn,
Commercial register: Amtsgericht Muenchen, HRB 153243,
Managing Directors: Charles Cachera, Michael Cunningham, Michael O'Neill, Eric Shander