ManageIQ/manageiq-providers-vmware

Adam Grare

@agrare

if they are running, and they aren't so backed up that they aren't getting to that queue item in 30 mins, then there's some other reason they aren't dequeueing messages

those are all the normal reasons

Fabien Dupont

@fdupont-redhat

Hi. Do you know if ManageIQ collect the history of metrics when enabling C&U or if the metrics will start from the date of activation ?

Adam Grare

@agrare

@fdupont-redhat there is a setting that controlls that, Settings.performance.history.initial_capture_days

by default it is 0

Fabien Dupont

@fdupont-redhat

@agrare thanks. Do you know what's the default history length for VMware ?

Adam Grare

@agrare

I believe it is configurable, trying to find out where the setting is

All values are in seconds. The default setting for vCenter Server is level 1, which retains sampled statistical data as follows:

5-minute samples for the past day
30-minute samples for the past week
2-hour samples for the past month
1-day samples for the past year

Adam Grare

@agrare

for historical we collect hourly so should be kept for a month by default

(totally untested though :wink: )

Fabien Dupont

@fdupont-redhat

So, in theory, we could have right-size recommendations from day 1, right ? That's good for migration :)

Unrelated question: who knows the AnsibleRunner implementation ?

Fabien Dupont

@fdupont-redhat

Probably the wrong room to ask...

Adam Grare

@agrare

Ladas and/or myself

but yeah, wrong room :)

Fabien Dupont

@fdupont-redhat

Hi. Any idea why a VM's location would be file:///[]/vmfs/volumes/5af94d9b-8e85fe68-9b02-b8ca3a6382e0/100_VM_TEST_15/100_VM_TEST_15.vmx, instead of just 100_VM_TEST_15/100_VM_TEST_15.vmx ? Have you ever met that ?

Adam Grare

@agrare

@fdupont-redhat that's the VM's summary.url, looks like it doesn't match any of the format's that we're expecting

https://github.com/ManageIQ/manageiq-providers-vmware/blob/master/app/models/manageiq/providers/vmware/infra_manager/refresh_parser.rb#L1510-L1542

what version of VC is that?

Daniel Berger

@djberg96

wow that is....eeesh

Adam Grare

@agrare

yeah, its pretty gross

personally I'd like to just return the storage.url instead of all that crap

Fabien Dupont

@fdupont-redhat

@agrare it appears with V2V and is grabbed by https://github.com/ManageIQ/manageiq/blob/master/app/models/service_template_transformation_plan_task.rb#L308 when building the URL for SSH transport method. It's what QE get in their lab (shutdown at the moment) when calling source.location. And source is a VMwrae VM.

Daniel Berger

@djberg96

@agrare was trying to create a uri template using addressable, but i cannot get around the guid being misinterpreted as a port

Adam Grare

@agrare

What are you trying to use that for? Local path on the host or something?

We'd need access to that environment to see what is going on

What version is that VC and those esxi hosts

Daniel Berger

@djberg96

was just trying to generalize the various filesystem uri's

e.g. Addressable::Template.new("{fs}://{dir}{/subdir}/{guid}")

Adam Grare

@agrare

I don't think that's generally applicable

There are so many different formats

Daniel Berger

@djberg96

right it would have to be more flexible, but maybe it's not the right approach

Adam Grare

@agrare

Yeah that's why I'm trying to understand how you want to use it

Daniel Berger

@djberg96

just trying to get rid of those explicit regexen, trying to make it friendlier

Adam Grare

@agrare

If you're trying to build something just for information we can do that

In the VMware refresh parser or in v2v?

Daniel Berger

@djberg96

vmware

(though perhaps v2v could use such a thing somewhere down the road)

Adam Grare

@agrare

What are you trying to use it for in v2v?

Daniel Berger

@djberg96

i don't have anything in mind atm, just tinkering

anyway, didn't mean to derail you

Adam Grare

@agrare

Okay let's not change the VMware parser without a good reason

It's gross but it handles a lot of different VMware VC and host versions

If you just want a way to identify the vm try using the uid_ems or ems_ref

Fabien Dupont

@fdupont-redhat

@agrare yes, that's what it's been used for since 5.9.4 :) Basically, the URI for SSH is in the form of ssh://root@10.0.0.1/vmfs/volumes/my_datastore/my_vm/my_vm.vmx. We could of course check if the source.location starts with file:// and get the end of the path only. But, I'd like to understand if that's something coming from VMware or ManageIQ.

Adam Grare

@agrare

Okay but you're not actually expecting to use that URI?

That is the URL from VMware

Fabien Dupont

@fdupont-redhat

Well, we're using the ssh:// URL. More precisely, virt-v2v consumes it to stream the disks. So, if VMware is sending weird data, we need to sanitize it. Should it be sanitized by inventory refresher or by V2V ?

Adam Grare

@agrare

Can we set up a meeting to discuss? And we'll need the source environment available

I'm still not sure how the vm location helps you with the disks since the disks aren't guaranteed to be in the same dir as the vm

Where communities thrive

People

Repo info

Activity