Created attachment 865537 [details] engine.log, both vdsm.log(s) Description of problem: DC is non-responsive and there's no SPM for 2 minutes when 3.3-based SPM host is put in maintenance in 3.2 environment. So... I have 3.2 env, I added 3.3 host into this 3.2 environment. This 3.3 host becomes SPM. But while putting this 3.3 SPM host into maintenance, SPM is not switched immediately but it takes 2 minutes, during that time DC is in Non-Responsive status (one cannot do anything). But... If I put 3.2 host into SPM, SPM is switched to 3.3 host in couple of _seconds_! 1. scenario: - 3.3 SPM -> Maintenance: 14:31:12 - 3.2 becomes SPM: 14:34:33 2. scenario: - 3.2 SPM -> Maintenance: 14:49:48 - 3.3 becomes SPM: 14:49:50 Version-Release number of selected component (if applicable): rhevm-3.2.5-0.49.el6ev.noarch 3.2 host: vdsm-4.10.2-30.1.el6ev 3.3 host: vdsm-4.13.2-0.11.el6ev How reproducible: 100% Steps to Reproduce: 1. have 3.2 setup with one 3.2 host and one 3.3 host 2. put 3.3 host into maintenance 3. wait how long does it take to make 3.2 SPM 4. 3.2 is SPM, 3.3 is active (again) 5. put 3.2 into maintenance 6. wait how long does it take to make 3.3 SPM Actual results: - 1st scenario cca 2 minutes - 2nd scenario cca 2 seconds Expected results: 2 seconds or less in both cases ;) Additional info:
fyi DC was iSCSI.
Following my discussion with Allon and Sean, as 3.2 engine is not supported in 3.5 version we are not going to test it.