Veeam One REporter: VM uptime/downtime issue

[Dietervdb]

Hello everyone,

We had a downtime on Veeam One Monitor server. You should think that every vm had the same uptime, but that is not the case.
For example:

Vm's in SLA service critical have al an uptime of 95,20%
VM's in SLA service important have all an uptime of 95,16%
VM's in SLA service standaard have all an uptime of 95,15%

Are VM's in different SLA groups polled on different ways or is there something else going on?

[Shestakov]

Hello Dieter,
If you have large infrastructure with lots of VMs and/or hypervisor doesn`t respond instantly, you can have a tiny shift between VMs data collection start time.
What interval have you set for the VM Uptime report? Trying to seize if 0.01-0.05% is a big deal in that case.
Thanks!

[Dietervdb]

That was indeed the solution, thanks.
I took an report of the service critical vm's and after a while I took an report of the service important vm's. I choose in both cases for "last month". When I choose from 1 november to 30 november in both cases I get the expected values.

I got also another question. I monitor 2 vcenters, and got an downtime of the veeam one server. I receive slightly different uptime values for the vm's on both vcenters. Can you explain why this is happening.

[Shestakov]

Let me briefly explain the mechanism. Once Veeam One service is started it tries to connect to VCs and VBR servers and once connection is established it collects real-time performance data from VCs every 20 seconds (Typical Mode). We use sophisticated technique with several metrics to determine if VM is up or down. During the time Veeam ONE server was down, data collection didn`t happen and VMs will be count as down in the report.
Could you specify the uptime difference, please?
It can be a result of connection difference between Veeam ONE server - VC1 and Veeam ONE server - VC2
Thanks.

[Vitaliy S.]

I monitor 2 vcenters, and got an downtime of the veeam one server. I receive slightly different uptime values for the vm's on both vcenters. Can you explain why this is happening.

Do you see slightly different uptime values only for the time period when Veeam ONE server was down? What about time periods before and after this downtime? What data do you see?

[Dietervdb]

@ Shestakov
The uptime for all the vm's of the first vcenter is 95,29% and for the second vcenter 95,32%. So its a tiny difference of 0,03% but is an difference of approximately 12-13 minutes on monthly basis. Can this be the case because of connectivity differences?

@Vitaliy S.
The downtime was on 15 November. If I took an report from 1 November to 14 November or from 16 to 30 November, I receive uptime values of 100%.

[Dietervdb]

I got also another question with this downtime from Veeam One Reporter:

All graphs of "network usage" have a gap when the downtime happens, what I think is normal. But graphs that show CPU usage or memory usage don't have this gap. The graph that shows CPU ready have sometimes this gap.

Is this a normal behavior?

[Vitaliy S.]

All performance metrics are collected by Veeam ONE Monitor and graphs in Monitor Client and Reports should be similar in terms of potential "gaps" and performance charts. The only difference is that reports are using 2 hours data samples in its charts, so seems like one of the data points was missing. Can you post a sample screenshot for this?

[Shestakov]

Thanks for the reply, Dieter,

The uptime for all the vm's of the first vcenter is 95,29% and for the second vcenter 95,32%. So its a tiny difference of 0,03% but is an difference of approximately 12-13 minutes on monthly basis. Can this be the case because of connectivity differences?

That`s why I asked about the reporting interval, 0.03% is around 20sec of a day, but 12min of a week. Also keep in mind 0.01% rounding inaccuracy. Anyway, even 8 minutes looks too much for the single connectivity, if only you didn`t have a lot of connection problems.
It`s just my educated guess, if the difference disturbs you, you can contact the support team to investigate the logs.
Thanks!

[Dietervdb]

@Vitaliy

Screenshot of cpu usage with no gap:


screenshot of network usage with a gap:


As you see, I dont understand why downtime causes a gap on the second graph and not also the first one.

[Vitaliy S.]

Ok, thanks for the screenshot. CPU and Network usage should return the same graphs. Can you ask our support engineer to take a look a the database and determine the reason for this discrepancy?

[Dietervdb]

Hi,

I'm monitoring 2 VMware data centers using a VM with Veeam One Monitor 8 and a separate VM with the MSSQL server for Veeam. Both are running on the same HP MSA storage.
Last month a controller on this MSA storage failed and needed to be rebooted. In total the storage was unavailable for around 30 minutes, but both the VM's survived.

When I take an up time report, the VMs of the first datacenter showed 99,17% up time, while the VMs of the other datacenter shows 100%.

This 99,17% means that this was an downtime of approximately 6 hours, while the storage was unavailable for around 30 minutes.

I think the reason is that up time data is collected in Veeam One Monitor for around 6 hours and is then writen to the database. I think I had bad luck that the data of the first data center was going to be written to the database during this failure of the MSA storage and thus creates this gap.

Is my story true or is there another reason?

[Vitaliy S.]

This could be a reason indeed as per our earlier discussion. As for the gaps showing up in the report, did you have a chance to clarify that with the engineer?