Web console is unresponsive and slow in larger domains (9 hosts, approximately 80 servers). Server listing operation (e.g. browsing though topology view) seem to be a main issue, however console in general is affected. Requests again http api on :9990/management are a lot faster than console, thus the issue is probably in the console itself.
This might be a significant issue for our customers, as some of them have even larger domains.
Harald Pehl <hpehl> made a comment on jira HAL-295 Several options here to improve the performance of the topology view: - Dedicated operation to read the topology - Reduce the payload to contain only those attributes which are actually displayed in the topology view - Consider paging: Issue request only for those hosts / servers visible on the relevant page
Harald Pehl <hpehl> made a comment on jira HAL-295 I can think of several options to improve the performance of the topology view: - Dedicated operation to read the topology - Reduce the payload to contain only those attributes which are actually displayed in the topology view - Consider paging: Issue request only for those hosts / servers visible on the relevant page
Beyond a performance issue, I think this is also a usability issue when you begin to have large domains for people to be able to manage it in the current layout. I have messaged the UXD team to see if they can help with a resolution that will meet both interaction & performance concerns.
Harald Pehl <hpehl> made a comment on jira HAL-295 I can think of several options to improve the performance of the topology view: - Dedicated operation to read the topology - Combine independent operations into one composite operation - Reduce the payload to contain only those attributes which are actually displayed in the topology view - Consider paging: Execute requests only for those hosts / servers visible on the relevant page
Do you still have the test setup with the big topology in place? If so I can provide you with a snapshot version of the console which includes an optimized topology view.
Harald Pehl <hpehl> updated the status of jira HAL-295 to Coding In Progress
Harald: We can create the environment whenever there are free machines. I will have a look on Monday when I am in the office.
Harald Pehl <hpehl> made a comment on jira HAL-295 Paging over the topology should use the in-memory representation of the topology, rather than reload the topology and start at another host index.
Harald Pehl <hpehl> updated the status of jira HAL-295 to Resolved
Harald Pehl <hpehl> made a comment on jira HAL-295 Closing this now as it will become part of WF 8.0.0.Final. Optimizations for reading the topology are in place. If it turns out that this is not enough, we can reopen / create a new issue targeting HAL 2.2
Reading the description in [1] a back port to 6.1.1 would be very complicated. I strongly recommend to update to 6.2 which includes optimizations in the console for large domains. [1] https://access.redhat.com/support/cases/01002475/
This was previously verified -- seems like the issue got stuck in the process and never reached QE... The impact of introduced optimization is noticeable.
Hi there, do we have any figure in terms of servers / HC supported in eap > 6.3 and eap 7? which size of domain have we tested, with no impact on performance?