Hello,

Yes, actually that is the case. /var/lib/nova/ partition hosts the machines. we have aound 170 GB on the partition. It seems to be the case the the value reported is the actual one, because there are thin-provisioned machines adding up to 420GB, which explains what is being reported to the monitoring.

I think this ticket can be closed, as the value is being reported by our backend and it is in fact correct. We will run an internal research to find out where the overprovisioning fits into the picture, as it doesn't seem to be configured in our system by default

Any responses? Can we close the ticket?

It would be nice if you provided the output of "df -f" command for any of your compute nodes.

on /var/lib/nova/instances if I'm not wrong. Can you confirm Atos Spain Tenerife Node Support Team ?

Where are those 400Gb disk mounted in the compute-nodes?

Ok I agree, anyway I just need the confirmation that the numbers about disk space from the command "nova host-describe" we analysed are dependant from that value. So maybe you could:

1) Launch the command "nova host-describe NODEx"
2) Change the replication factor value and restart nova
3) Launch again the command "nova host-describe NODEx"

see if those numbers changes and give me confirmation.

Then we can close this ticket and I will plan the implementation of a new feature in the monitoring system in order to support this disk replication factor. Once ready we could test and tune the value for your region.

Sounds good ?

Hi, Daniele

Yes, I have been checking this point. In our nova.conf configuration files, I haven't found the disk_allocation_ratio parameter, which makes me think either default value is not 1, or the problem lies somewhere else.

In order to push this forward, this is my proposal: I am going to force 1,5 as disk_allocation_ratio on the controllers, and then you can patch the monitoring observing that actually the value could be different than 1. That way we can check if the values match afterwards

I did some inspection on the monitoring API code and it confirms that we do not consider “disk_allocation_ratio”, or better, we consider 1 as the default. Therefore if you have a different value set for your region the disk space reported by the monitoring API is wrong.

Atos Spain Tenerife Node Support Team: Have you had the chance to check which is the value of that configuration parameter on your region ? As specified in table 3.52 here: https://goo.gl/5WRrkf it should be present in the nova.conf file on your controllers.

I need this information in order to confirm the cause of the issue, then I can start the development of a new feature in the monitoring system in order to support this region customisation.

This issue is related with OPS-1445. Status can be tracked also from it.

Dear, can we try to go ahead on this ticket ? We have planned to solve it on the next Sprint 6.2.2. Please provide required information.

Dear operators, any news about my last request sent via mail ?

Monitoring chain has been fixed and task is back to progress state, I'm waiting detailed information from the node operators.

We currently have a problem on the monitoring chain which prevents Infographics to show latest monitored data from the regions. For this reason I cannot verify your problem right now but I will take a look once the issue with the monitoring system will be resolved.

Dear,

I will take a look at this issue and I will update you as soon as possible.

Daniele

Hi Daniele,

It seems Pablo forgot to assign the ticket to you. Could you kindly take a look? We would really like to sort this issue out soon.

Thanks in advance for your help.

Regards,

ilknur

Well,

This issue falls between 3 different components (and maintainers): Kasia Di Meo would be the right person for visualization issues, Daniele Santoro for wrong aggregated information at Monitoring API, and me for the rest of components of Monitoring.

I think that, in this case, we should assign the ticket to Daniele and he will contact the rest in order to solve this known problem which involves different components.

Regards.

Dear Pablo,

I am assigning this issue to you as I believe you are the maintainer of the Infographic monitoring tools. Could you kindly take a look if there is a defect in the tool, or a missing/incorrect configuration on our part, etc?

Thanks,

ilknur

The issue has been emailed:

• Time sent: 17/Nov/16 1:07 PM
• To: roberto.gil@atos.net
• with subject: *(HELP-7721) [Fiware-lab-help] SpainTenerife node - Incorrect disk ocupation values reported on Infographics *
  

Dear Roberto,

I forwarded your problem to the Spain Tenerife Support Team.

Best Regards,
Gergő