[Freeipa-users] Upgrade from 3x to 4x cant create first replica.

Martin Kosek mkosek at redhat.com
Mon Feb 9 16:36:31 UTC 2015


On 02/09/2015 05:16 PM, Chris Mohler wrote:
> On 02/09/2015 10:18 AM, Martin Kosek wrote:
>> On 02/07/2015 12:27 AM, Chris Mohler wrote:
>>> I'm having some troubles. I have an older IPA install Version 3.0.0. on Centos
>>> 6.6. It's currently the only master for my domain. I have about 4k user
>>> accounts on here and it's a live system called "idm"
>>>
>>> I'm trying to upgrade to V4.x as I am hoping to fix some issues I am having.
>>> (clients can't auth unless service sssd is restarted multiple times "10 (User
>>> not known to the underlying authentication module") I think this is possibly
>>> unrelated and the topic for another thread.
>>>
>>> I created a new VM and installed Fedora Server 21 and FreeIPA 4.1.2 it's called
>>> "ipa"
>> Good. Also note that we RHEL/CentOS 7.1 will have FreeIPA 4.0+ version baked
>> in, so you can also use that platform if you are used to it.
>>
>>> on the master "idm" I ran "ipa-replica-prepare" and transfered the file to the
>>> future replica "ipa" Then I ran the install replica script ipa-replica-install
>>> --setup-ca /home/svradm/replica-info-ipa.cs.oberlin.edu.gpg
>>> Things went well until it failed
>>>
>>> [24/35]: setting up initial replication
>>> Starting replication, please wait until this has completed.
>>> Update in progress, 133 seconds elapsed
>>> Update in progress yet not in progress
>>>
>>> Update in progress yet not in progress
>>>
>>> Update in progress yet not in progress
>>>
>>> [idm.cs.oberlin.edu] reports: Update failed! Status: [10 Total update
>>> abortedLDAP error: Referral]
>>>
>>> [error] RuntimeError: Failed to start replication
>>>
>>> Your system may be partly configured.
>>> Run /usr/sbin/ipa-server-install --uninstall to clean up.
>>>
>>> Please help I'm getting nowhere by myself.
>> Can you please look on the master you are replicating from and look for errors
>> in /var/log/messages or DS errors log?
>>
>> Maybe you will see messages like "ns-slapd: encoded packet size too big (xxxxxx
>>> 65536)" that are know to pop up more with CentOS 6.6.
> Hi Martin,
> Thanks for the reply and help I appreciate it.
> 
>> Good. Also note that we RHEL/CentOS 7.1 will have FreeIPA 4.0+ version baked
>> in, so you can also use that platform if you are used to it.
> Good to know. I try to be distro agnostic. I've used Redhat 7.1 then went
> Solaris, then Ubuntu, Now I'm back for Centos and Fedora. I guess I'm equally
> uncomfortable with either version.
> 
> That Said. Is there any reason that I could or should not have a replica on a
> Fedora 21 server and 2nd replica on a Centos 7.1 later? My understanding is the
> more the merrier.

It should just work. Just note that in case of Fedora Server, these are
upstream/Fedora bits which are only tested upstream. So if you for example
break something in Fedora 21 (not likely to happen though ;-) and then get the
change *replicated* to RHEL production instance, I do not think Red Hat support
would be happy with that.

Also, if for example upstream releases FreeIPA 4.2, I would not just plug it in
your production RHEL instance is it would upgrade all the data for 4.2 level -
which should get more downstream testing before Red Hat can rubber stamp it.

TLDR; if you are happy with the upstream level of support (this list/IRC/Trac),
knock yourself out :-)

>> Can you please look on the master you are replicating from and look for errors
>> in /var/log/messages or DS errors log?
> 
> I tried to setup the replica again just now so I have some fresh logs.
> 
> From the Dirserv error log
> [08/Feb/2015:22:14:48 -0500] - 389-Directory/1.2.11.15 B2014.314.1342 starting up
> [08/Feb/2015:22:14:48 -0500] schema-compat-plugin - warning: no entries set up
> under cn=computers, cn=compat,dc=cs,dc=oberlin,dc=edu
> [08/Feb/2015:22:14:50 -0500] - slapd started.  Listening on All Interfaces port
> 389 for LDAP requests
> [08/Feb/2015:22:14:50 -0500] - Listening on All Interfaces port 636 for LDAPS
> requests
> [08/Feb/2015:22:14:50 -0500] - Listening on
> /var/run/slapd-CS-OBERLIN-EDU.socket for LDAPI requests
> [09/Feb/2015:10:40:30 -0500] NSMMReplicationPlugin -
> agmt="cn=meToipa.cs.oberlin.edu" (ipa:389): Schema replication update failed:
> Server is unwilling to perform
> [09/Feb/2015:10:40:30 -0500] NSMMReplicationPlugin - Warning: unable to
> replicate schema to host ipa.cs.oberlin.edu, port 389. Continuing with total
> update session.
> [09/Feb/2015:10:40:30 -0500] NSMMReplicationPlugin - Beginning total update of
> replica "agmt="cn=meToipa.cs.oberlin.edu" (ipa:389)"
> 
> To be fair and not duplicate efforts I have had the following error
> [08/Feb/2015:08:51:26 -0500] - WARNING: userRoot: entry cache size 10485760B is
> less than db size 12115968B; We recommend to increase the
> entry cache size nsslapd-cachememsize.
> 
> To which I have asked another question "how do I change the entry cache size"
> https://www.redhat.com/archives/freeipa-users/2015-February/msg00114.html
> I now get additional errors which I would guess are possibly related.

IMO, they this should not be related (should not break replication). I do not
see anything useful in the error log though. Did you also check
/var/log/messages for the errors log I sent?




More information about the Freeipa-users mailing list