<html> <head> <meta content="text/html; charset=ISO-8859-1" http-equiv="Content-Type"> </head> <body bgcolor="#FFFFFF" text="#000000"> <div class="moz-cite-prefix">On 10/29/2015 01:28 PM, thierry bordaz wrote: </div> <blockquote cite="mid:563210E0.4060607@redhat.com" type="cite"> <meta content="text/html; charset=ISO-8859-1" http-equiv="Content-Type"> <div class="moz-cite-prefix">On 10/23/2015 10:44 AM, Ludwig Krispenz wrote: </div> <blockquote cite="mid:5629F352.9050005@redhat.com" type="cite">Hi, the attached two patches address issues I found when testing ca management in the topology plugin Thanks for review, Ludwig <fieldset class="mimeAttachmentHeader"></fieldset> </blockquote> Hi Ludwig, Patch 20 is good to me. I have one remark, you call ipa_topo_cfg_host_find with lock flag. So that the replica config is not updated during the test. Now the lock protects each call separately. The risk is very low that the target host could become unmanaged by the time we test the source host. </blockquote> yes, and if two paralle operations do related things like adding an agreement and making a host managed/unmanaged there is a race for the lock. The lock itself cannot prevent these things, it only can protect the data structures from being read while modified. Also with two separate locked calls the second call has a chance to be aware of parallel changes <blockquote cite="mid:563210E0.4060607@redhat.com" type="cite"> ACK. Patch 21 is also good. Just in ipa_topo_util_init_hosts, why not calling ipa_topo_cfg_host_add to not duplicate the source ? </blockquote> no reason, revised patch is attached, thanks for noticing <blockquote cite="mid:563210E0.4060607@redhat.com" type="cite"> thanks thierry </blockquote> </body> </html>