Description of problem: - When deleting source VM during import, the import progress get stuck in webadmin. - Observing VdsStats shows that v2v job is still running. - Observing host processes shows that there are no v2v processes. Version-Release number of selected component (if applicable): rhevm-3.6.2-0.1.el6 sanlock-3.2.4-1.el7.x86_64 libvirt-client-1.2.17-13.el7_2.2.x86_64 qemu-kvm-rhev-2.3.0-31.el7_2.4.x86_64 vdsm-4.17.15-0.el7ev.noarch How reproducible: Consistently. not dependent in import progress phase (initializing, copying etc.). Steps to Reproduce: 1. Import VM from VMware environment. 2. During import, delete VMware VM. 3. Actual results: Import process get stuck without rolling back. Expected results: VM conversion should be canceled with rollback. Additional info: vdsm and engine log attached. Import started at 2016-01-13 09:02:10,873 (engine.log)
Created attachment 1114355 [details] engine log
Created attachment 1114356 [details] vdsm log
Please add logs/proof that on host v2v process was no longer running; or v2v logs. Looking at vdsm logs there is no indication about anything - Shahar, are you sure when v2v dies it is reported back to vdsm correctly?
(In reply to Michal Skrivanek from comment #3) > Please add logs/proof that on host v2v process was no longer running; or v2v > logs. > > Looking at vdsm logs there is no indication about anything - Shahar, are you > sure when v2v dies it is reported back to vdsm correctly? Yes, as a test I started to import and stop the engine (for not rolling back and clear the job) killed virt-v2v and the stats show: v2vJobs = {'423c38b0-37f6-4d1b-da17-bd1d55fe296c': {'description': "Job '423c38b0-37f6-4d1b-da17-bd1d55fe296c' process failed exit-code: -9, stderr: ", 'progress': 0, 'status': 'error'}}
so, that would be an indication that v2v didn't die, right (adding back needinfo on nisim)
I reproduced it again.. when deleting source VM during rhevm "copying disk" phase, the import get stuck in this phase forever. as for the jobs and processes: v2v job is still running (see getVdsStats.txt attached) v2v process is also running [virt-v2v] <defunct> (see host_process.txt attached) Also, see vmware_vms.txt for existing vmware VMs. the source VM deleted name is RHEL7_test. vdsm and engine log also attached. Another thing important to mention, when deleting source VM during initialize phase, import progress is rolled back as expected.
Created attachment 1120441 [details] VMware VMs
Created attachment 1120442 [details] host processes
Created attachment 1120443 [details] getVdsStats
Created attachment 1120444 [details] new vdsm.log
Created attachment 1120445 [details] new engine.log
Verified. rhevm-3.6.6.2-0.1.el6 qemu-kvm-rhev-2.3.0-31.el7_2.10.x86_64 vdsm-4.17.28-0.el7ev.noarch sanlock-3.2.4-2.el7_2.x86_64 libvirt-client-1.2.17-13.el7_2.4.x86_64 Verification scenario: 1. Import VM from VMware. 2. During "initializing" phase, delete source VM and verify ConvertVmCallback started, VM deleted from webadmin and its disks were deleted also. 3. Import another VM from VMware. 4. During "copying" phase, delete source VM and verify ConvertVmCallback started, VM deleted from webadmin and its disks were deleted also.