Bug 890823
| Summary: | ovirt-engine-backend: Host 'load bar' do not count VMs in Status Paused. | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | Red Hat Enterprise Virtualization Manager | Reporter: | Omri Hochman <ohochman> | ||||||
| Component: | ovirt-engine | Assignee: | ofri <omasad> | ||||||
| Status: | CLOSED CURRENTRELEASE | QA Contact: | Jiri Belka <jbelka> | ||||||
| Severity: | medium | Docs Contact: | |||||||
| Priority: | medium | ||||||||
| Version: | 3.1.0 | CC: | dfediuck, dyasny, iheim, lpeer, pstehlik, Rhev-m-bugs, sgrinber, yeylon, ykaul | ||||||
| Target Milestone: | --- | ||||||||
| Target Release: | 3.2.0 | ||||||||
| Hardware: | x86_64 | ||||||||
| OS: | Linux | ||||||||
| Whiteboard: | sla | ||||||||
| Fixed In Version: | sf4 | Doc Type: | Bug Fix | ||||||
| Doc Text: | Story Points: | --- | |||||||
| Clone Of: | Environment: | ||||||||
| Last Closed: | Type: | Bug | |||||||
| Regression: | --- | Mount Type: | --- | ||||||
| Documentation: | --- | CRM: | |||||||
| Verified Versions: | Category: | --- | |||||||
| oVirt Team: | SLA | RHEL 7.3 requirements from Atomic Host: | |||||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||||
| Embargoed: | |||||||||
| Attachments: |
|
||||||||
|
Description
Omri Hochman
2012-12-30 13:39:13 UTC
Created attachment 670415 [details]
pasued_vms
Created attachment 670416 [details]
engine.log
Omri, the load bar was recently removed, as we discovered it has nothing to do with workload and it was simply counting VMs, so a strong host running 90 VMs is considered 'loaded' even if it's capable of running 200 VMs. We do leave the count of running VMs, so the actual representation of host load will now be represented by 3 resource load bars (cpu, memory and network). So in this context, I'd like to verify the running VMs will actually represent running VMs only, and not VMs in paused state. Do you see any good reason why 'running' VMs should include the pause state? Remember it's only a number now, and has no bar and / or colors. I think that As a 'user' when you see 0 VMs on host, you may think its 'OK' to switch the host to maintenance, while if there are paused VM's on it, maintenance will fail (Host will remain in 'Preparing For Maintenance'). Other thing worth mentioning is that the "common" scenario of many VMs in 'paused' status, can occur when VMs attempt to write to storage and getting IO Error, i.e, storage problem or no storage space left, in those cases you can have an entire environment running paused VMs and the load bar won't reflect it. RHEVM 3.2 (build SF2.1) the load bar still there. Thanks Omri, makes sense. Simon, how do you see this issue? Today we simply check the vm_active report we get from vdsm, which will give us the current number. In theory, we can go over all vm's for each host and give better report, but this may cause performance issues. But if we can find a way, which status would you like to see in this column? (In reply to comment #5) > Thanks Omri, > makes sense. > > Simon, how do you see this issue? A paused VM still consumes Memory so it can't be ignored. There is not much difference between this VM and any other VM that is 99% idle, actually in some cases it's loads the host more since it may consume more memory. Load bar (and maybe we should call in VM count bar) should show all the VMs on the host regardless of status. I actually have this exact scenario on my setup where I lost one storage domain. I have 3 paused and one running VMs And if you'll look at the VDSM stats you can see: vmActive = 1 vmCount = 4 You are reporting the active, but should report the total count. This means the change should not be complicated. Very well, makes sense. So column title will change to "VMs", and the figure will be taken from VDSM report of vmCount. (In reply to comment #7) > Very well, makes sense. > > So column title will change to "VMs", and the figure will be taken from > VDSM report of vmCount. Ack, Just try not to use abbreviations in column headers or do it carefully. It's either "Virtual Machines" or something similar. posted upstream: http://gerrit.ovirt.org/#/c/10968/ ok, sf5, 1 active and 1 paused = in Hosts|Virtual machines column, the number is '2'. OK. 3.2 has been released 3.2 has been released 3.2 has been released |