The "terminate instances" button sometimes does not appear when there are many instances (> 200)

[Franck Dernoncourt]

The "terminate instances" button sometimes does not appear when there are many instances (> 200), even after checking some checkboxes next to instances' names on the left. Demo: http://youtu.be/QDlyV82shDc

I have noticed this behavior on two browsers (Chrome/Firefox) and two OS (Windows/Ubuntu).

[Julie Pichon (jpichon)]

This is strange. From the video it looks like the 'Terminate' button is still available on individual rows. Does it happen consistently once it starts happening, or does the button comes back after a refresh? If you could look through the logs when this happens, there might be a hint to give us more information as to where it is failing exactly. Thanks for the bug report.

[Franck Dernoncourt (franck-dernoncourt)]

Pretty consistently: it seems like the higher the number of nodes, the more often the button disappears (~softmax proba style). What log are you referring to? Google Chrome logs?

[Julie Pichon (jpichon)]

I suspect this may be due to a problem in the Horizon code itself, so if you have access to the server logs (in /var/log/httpd, /var/log/apache2 and/or /var/log/horizon) that would be helpful. I'll see if I can find an environment to try and reproduce it as well.

[Franck Dernoncourt (franck-dernoncourt)]

Sorry I don't have access to the server logs (not the admin): the admin told me he suspects that the Terminate Instances button is timing out due to the large number of instances in the project.

[kennychu (kenny-chu123)]

I also had meet this issue. And volume table "Delete Volumes" sometimes also disappeared.
I trace the horizon/tables/actions.py BatchAction function.

    def _allowed(self, request, datum=None):
        # Override the default internal action method to prevent batch
        # actions from appearing on tables with no data.
        if not self.table.data and not datum:
            return False
        return super(BatchAction, self)._allowed(request, datum)

Sometimes the self.table.data was None. So the button will disappear if we hit this situation.
Does anyone have good idea to deal with it?

[Gloria Gu (gloria-gu)]

I tried it...looks like when click more to go to next page ...one the next page, the table "Terminate instance" button is gone. row's is still there. I will take a further look and give it a try.

[kennychu (kenny-chu123)]

My procedure is launch many instances.
And the first instance appears process bar in task field may hit the issue.
Create volumes also hit the issue.

The table action "Terminate Instances" "Soft Reboot Instances" and "Delete Volumes" have the "btn-danger " class.
May be it's related horizon/statics/ js/horizon/horizon.tables.js file.

[Gloria Gu (gloria-gu)]

It only happens when still creating large amount of instances....once they are done...the button shows up on every page...

[Gloria Gu (gloria-gu)]

my guessing...it might be related to some racing condition...when large amount of data is processed at same time...self.table.data is not updated quickly. I don't know how to fix it yet, will dig more for one more day...otherwise will release this bug

[Gloria Gu (gloria-gu)]

rethink about the bug... based on my situation

If it is still creating a large amount of instances...while it is still creating, it might be a "Good" idea, not showing the "Terminate Instance" button. I think the first page it shows that button maybe because it has a couple of instances already created while other are still creating..... Meanwhile, on the next page all the instances are still creating so it should not show "Terminate instance" button.

Make sense?

[Gloria Gu (gloria-gu)]

I could not reproduce the same situation as the video shows. Once all the instances settle...the button shows up ok for me.

releasing the bug for other folks to take a look.

[Gloria Gu (gloria-gu)]

While I am reviewing the following code...I tried this bug again with the following changes

https://review.openstack.org/#/c/113855/1

noticed the problem I have seems to go away. However the fix has other issues why I try to delete instances after everything settled...not sure the issued caused by the fix though.

[kennychu (kenny-chu123)]

I tried the bug and following the changes
https://review.openstack.org/#/c/113855/1
the bug was still existed.

[Doug Fish (drfish)]

I've recently been made aware of an occurrence of this bug. It was seen on an Juno installation. Unlike some of the other reports there was a modest number of instances available (maybe 10 or 15?), however 2 of them were stuck in a transitional state, resulting in loads of requests coming in as described in bug https://bugs.launchpad.net/horizon/+bug/1263665

Maybe the missing button is somehow related to the large number of refreshes, or the resulting poor performance?

[Doug Fish (drfish)]

oh - also I saw the "Soft Reboot Instances" button should have been shown as well, but was missing.

[Bhakti Panda (bhakti.panda)]

As no one is working on this. i am assigning it to myself. Investigation in progress.....

[Akihiro Motoki (amotoki)]

The bug was reported four years ago. Does this still happen in the recent releases?

[Franck Dernoncourt (franck-dernoncourt)]

I don't know

On Sat, Dec 29, 2018, 21:04 Akihiro Motoki <<email address hidden>
wrote:

> The bug was reported four years ago. Does this still happen in the
> recent releases?
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1313202
>
> Title:
> The "terminate instances" button sometimes does not appear when there
> are many instances (> 200)
>
> Status in OpenStack Dashboard (Horizon):
> Confirmed
>
> Bug description:
> The "terminate instances" button sometimes does not appear when there
> are many instances (> 200), even after checking some checkboxes next
> to instances' names on the left. Demo: http://youtu.be/QDlyV82shDc
>
> I have noticed this behavior on two browsers (Chrome/Firefox) and two
> OS (Windows/Ubuntu).
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/horizon/+bug/1313202/+subscriptions
>

[Launchpad Janitor (janitor)]

[Expired for OpenStack Dashboard (Horizon) because there has been no activity for 60 days.]