Fact obtained from puppet is not reflected in Racks location data

Bug #1558040 reported by Igor Shishkin
10
This bug affects 2 people
Affects Status Importance Assigned to Milestone
Fuel for OpenStack
Invalid
Medium
Fuel Infra Apps
Mitaka
Won't Fix
Medium
Fuel Infra Apps
Newton
Invalid
Medium
Fuel Infra Apps

Bug Description

Hello,

https://inventory.infra.mirantis.net/650

Here's an example of srtext location obtained from puppet which is not reflected in object location.

Please fix.

Revision history for this message
Alexander Charykov (acharykov) wrote :

It is reflected on raw page. Facter location shown on main page sets manually.

Changed in fuel:
status: New → Invalid
Revision history for this message
Igor Shishkin (teran) wrote :

Anyway it breaks previous behaviour and make us suffering with being forced to set locations manually.
I believe this should be address or replaced with other task.

Changed in fuel:
status: Invalid → New
Revision history for this message
Alexander Charykov (acharykov) wrote :

Fact is available in raw list. It is not a bug, it is intensional, that location fact is not automatically updates facter_location on InventoryObject.

Changed in fuel:
status: New → Invalid
Revision history for this message
Igor Shishkin (teran) wrote :

@Alexander, the thing is not about raw list, the thing is about location property.
Now we're forced to set up location manually, which is not such a good idea because amount of servers we have.

So please fix the issue or suggest another way how to achieve the same - i.e. get locations set automatically(technically it could be possible to do from DCIM, btw).

Changed in fuel:
status: Invalid → New
Revision history for this message
Alexander Charykov (acharykov) wrote :

Fact location should not be reflected in Racks location data, because source of this fact is racks itself.

PXEtool gets those fact from racks and pass it to host.

So it is not a bug. Please create request to set location based on DCIM entry.

Changed in fuel:
status: New → Invalid
Revision history for this message
Igor Shishkin (teran) wrote :

Since it's broken functionality it's a bug at the moment so we should keep previous behaviour or switch to use DCIM one, since current functionality is unexpected(means broken).

Changed in fuel:
status: Invalid → New
Changed in fuel:
status: New → Invalid
Revision history for this message
Igor Shishkin (teran) wrote :

@Alexander, please always put comments about status changes.

Changed in fuel:
status: Invalid → New
Revision history for this message
Alexander Charykov (acharykov) wrote :

@Igor I've already wrote why bug is invalid. Do not change status back.

Changed in fuel:
status: New → Invalid
Revision history for this message
Igor Shishkin (teran) wrote :

@Alexander, you have forced us to work with locations manually at moment without any notification or workaround.
It's unacceptable and not possible for new orders while we don't even have bulk actions.
I believe this is broken functionality until the way of using locations feature won't be provided in useful way.

Changed in fuel:
status: Invalid → New
Revision history for this message
Alexander Charykov (acharykov) wrote :

@Igor it is not "broken functionality" see my comments #3 and #5. It was working incorrectly, so it was fixed.

If you want bulk actions please create a bug for it.

Changed in fuel:
status: New → Invalid
Revision history for this message
Igor Shishkin (teran) wrote :

Alexander, the thing is puppet is the only place knows about location for both of VMs and HWs, so I suppose if you don't want to hardcode that binding let's make it configurable.
But now previous behavior we had when PROD was completed is broken, I'm not talking it was right behavior was racks was accepted with it. So previous and useful behavior broken.

Changed in fuel:
status: Invalid → New
Revision history for this message
Alexander Charykov (acharykov) wrote :

Igor, puppet isn't the place where is location is real location of object. In case if someone, set it to incorrect value, host can move to another location or not to any location at all.
Also such behavior reflects on pxetool, which would remove host from location, and add new host to another location.

Such changes requires confirmation from engineer, however it takes too much time, to update location for every because we don't have mass confirm/update changes in racks.

Changed in fuel:
status: New → Invalid
Revision history for this message
Igor Shishkin (teran) wrote :

@Alexander, puppet is the only place knows about locations for both of VMs and HW nodes, so it's the best place to retrieve such kind of data automatically.
If we set wrong location it's representation in racks will become our smallest problem, so I don't think it makes sense to use this point.

Setting up the node could be truly treated as such kind of confirmation. So I believe this would be enough.
I'm almost sure we should not hardcode such behaviour but add a way to rely on puppet facts we should.

Changed in fuel:
status: Invalid → New
Revision history for this message
Alexander Charykov (acharykov) wrote :

See comments above: facter location is shown in raw, auto update will not be done.

Revision history for this message
Igor Shishkin (teran) wrote :
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.