rafaelweingartner commented on issue #1124: CID-1338387: Deletion of method endPointSelector.selectHypervisorHost URL: https://github.com/apache/cloudstack/pull/1124#issuecomment-329965501 I do not know what was(is?) broken. Can you explain some things for me? You say: > For example by default CopyCommand involving only image stores goes to SSVM but for some hypervisors there is some optimisation done and can be directly handled by hypervisor agent. With this method we can change that default behaviour. And the removed method was giving us that different endpoint. And you also said that this was reported in https://issues.apache.org/jira/browse/CLOUDSTACK-9025. Is the PR responsible to close the #CloudStack-9025 ticket, the #1030/#1176 that is linked to it? Reading that ticket: > As default, CloudStack create linked clone snapshot for VM instance . When we take a snapshot for the VM, and create a template based on such snapshot, CloudStack only download incremental VHD as template file, as a result, the VM instance fail to deploy as it is incomplete. For me, the ticket you mention does not seem to be related to the optimization you are talking about. If I only consider the ticket information, nothing should be working right (by that ticket VMs should not start without that fix)? You said that this PR removed the code introduced to solve the aforementioned ticket. The interesting part is that we did not remove the code introduced in #1030/#1176, the code is still there. We are only not using a method that was and is (at least for me) not making much sense. The removed method is not introduced in #1030/#1176. We only removed the method `endPointSelector.selectHypervisorHost(new ZoneScope(host.getDataCenterId()));` The use of this method to achieve the optimization you are talking about seems to rely on very strict and specific configuration for the system (or lucky ) to get things working, which is an approach that I personally prefer to avoid. But if more people are willing to support the re-introduction of this method, I cannot say much more against it. I would be inclined though to discuss the introduction of code to execute the optimization you mentioned in a more direct, flexible and clear way. BTW: I use, and I know quite some production systems that use multiple hypervisors types per zone, so CloudStack has to decide if it adversities that multiple are supported, but not per zone, then this has to be clear. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
With regards, Apache Git Services
