In the normal scenario, yes, all segments are deleted. But in scearios, where you remove a central node from a topology, perhaps even before removing it, the distance applied to a node cannot be replicated to other separate nodes. We can make the maximum effort, not well enough at the moment, but no guarantee that everyone will be cured automatically. GitHub is home to more than 50 million developers who work together to host and verify code, manage projects and create software together. Is it possible that I am missing something, but should we not remove all the agreements that link to a replica if we remove a replica? The situation described in the previous remark would never happen. As soon as we remove B, the adaptation agreements will also be removed. Topology plugin has been moved to the next short version – 4.3, so troubleshooting and checking this ticket can wait. Isn`t that fixable with: `ipa-replica-manage del –force $HOST 2`? If the connection between 1 and 2 was removed and host3, there would be no way to communicate on Host2 to Host1, so I don`t think we could do much. A replication agreement is already available for this host. It needs to be removed.
Run this on the master that generated the information file: % ipa-replica-manage del ipa2.quartzbio.com –force If you try to delete server B, you abandon A and C and leave the entry cn-masters because the replication is separate. I have no idea or advice. Could we put this on the Freeipa user mailing list? When deleting a separate IPA replication server that had replication agreements with multiple IPA hosts, the metadata of this replicant is still present on the IPA servers that are still in progress. This does not allow you to completely remove the separate IPA server. There are special tasks that allow you to delete this metadata. For example, I stopped the replicating docker. When I tried to run it again, I received this message: Yes, it can be done, but the problem is still that deleting a Masters may not be replicated on all servers. There are two different scenarios for breaking topology: Suppose a topology A B C would be the second most interesting, because vm-134 is most likely a straight knot. And so this research would have had nentries 1 But in 2] it can not work, there is no way to propagate operations on C until the topology is repaired C will have a bad condition, provided you have a topology associated with A, B, C and refuse any distance of segments.