Horizon incorrectly displays memory of an instance

[Ryan Aydelott]

Horizon incorrectly reports the size after VM's reach a size greater then 1TB.

For example openstack generated this xml for a 1.5TB memory VM:

<domain type="kvm">
 <uuid>3e3f022b-5891-48cf-a09b-ea3fb29e4006</uuid>
 <name>instance-0000444f</name>
 <memory>1584713728</memory>
 <vcpu>24</vcpu>

But viewing this in Horizon only reports the VM memory size @ 1TB (see attached screenshot)

[Gary W. Smith (gary-w-smith)]

It appears that your screenshot and the XML in the description are for different VMs. They have different names and IDs.

[Ryan Aydelott (ryade)]

ok - let's repeat:

UUID in the XML== <uuid>3e3f022b-5891-48cf-a09b-ea3fb29e4006</uuid>

The UUID in the screenshot is identical. Can you explain to me how you believe this is a different VM?

I've attached (another) screen cap of this browser window with the screen cap and this comment attached.

[Ryan Aydelott (ryade)]

Output of Nova-Show that correlates the UUID to instance_name:

root@u10-p:~# nova show 3e3f022b-5891-48cf-a09b-ea3fb29e4006
+--------------------------------------+----------------------------------------------------------+
| Property | Value |
+--------------------------------------+----------------------------------------------------------+
| status | ACTIVE |
| support_vl130 network | 10.1.0.3 |
| updated | 2014-10-22T20:29:05Z |
| OS-EXT-STS:task_state | None |
| OS-EXT-SRV-ATTR:host | cm13-p |
| key_name | None |
| image | Ubuntu-14.04 (f9b734fe-a413-4e46-ae7f-7c32a0728954) |
| hostId | 0b9c87282865c59532e1c3e99b40a1f26914c51c84b7c41fa4b9e8ec |
| OS-EXT-STS:vm_state | active |
| OS-EXT-SRV-ATTR:instance_name | instance-0000444f |
| OS-SRV-USG:launched_at | 2014-10-22T20:29:05.000000 |
| OS-EXT-SRV-ATTR:hypervisor_hostname | cm13-p |
| flavor | mem2.100 (327) |
| id | 3e3f022b-5891-48cf-a09b-ea3fb29e4006 |
| security_groups | [{u'name': u'default'}] |
| OS-SRV-USG:terminated_at | None |
| user_id | 5ef4ddd7c64545eda6a37cbd0da77c35 |
| name | ryanmem2test |
| created | 2014-10-22T20:28:24Z |
| tenant_id | 030d99a0fd014dd1859a1f633822ce29 |
| OS-DCF:diskConfig | MANUAL |
| metadata | {} |
| os-extended-volumes:volumes_attached | [] |
| accessIPv4 | |
| accessIPv6 | |
| progress | 0 |
| OS-EXT-STS:power_state | 1 |
| OS-EXT-AZ:availability_zone | nova ...

[Gary W. Smith (gary-w-smith)]

Sorry, the id's were the same. The instance name in your xml did not match the instance name in the screenshot.

[Ryan Aydelott (ryade)]

So there is name, and instance_name

| OS-EXT-SRV-ATTR:instance_name | instance-0000444f |
| name | ryanmem2test |

Name displayed in Horizon is exactly that: ryanmem2test
instance_name (as an attribute) is what appears in the xml: <name>instance-0000444f</name>

Either way, this doesn't invalidate the bug report, as Horizon does not accurately display the memory of instances >1tb.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/137593

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/137593
Committed: https://git.openstack.org/cgit/openstack/horizon/commit/?id=12f57f65a8572e981bd27676ee89b30e2202c07c
Submitter: Jenkins
Branch: master

commit 12f57f65a8572e981bd27676ee89b30e2202c07c
Author: Sam Betts <email address hidden>
Date: Thu Nov 27 10:35:06 2014 +0000

    Show up to 1 decimal place for instance detail RAM

    The current instance details page rounds the amount of RAM to the
    nearest integer, which at GB and TB levels does not provide enough
    fidelity to be useful, this patch allows the page to display up to one
    decimal place if it is required, by changing the filter applied in the
    template from mbformat to mb_float_format.

    Change-Id: Ida97e83285b64ac6b08260f78f31c69dc37009b0
    Closes-Bug: 1384446