[Murano] [Artifacts] Murano dashboard doesn't get Murano API url

[Alex Kholkin]

Steps to reproduce:
1. Deploy environment with Murano
2. Add experimental group and reset environment
3. Choose in settings 'Enable glance artifact repository' and redeploy env
4. Go to horizon and navigate to 'Murano'>'Environments' and try to create env

Expected results:
You will be able to create the env

Actual result:
Dashboard is saying that there is no connection to the Murano API

Additional information:
1. Adding "MURANO_API_URL = 'http://192.168.0.2:8082'" to the /etc/openstack-dashboard/local_settings.py will fix this problem

2. ISO 305

3. This is a part of the Horizon logs

2016-05-11 11:33:29,979 1348 WARNING horizon.exceptions Recoverable error: Error communicating with http://localhost:8082 HTTPConnectionPool(host='localhost', port=8082): Max retries exceeded with url: /v1
/catalog/packages/categories (Caused by NewConnectionError('<requests.packages.urllib3.connection.HTTPConnection object at 0x7fccd9df49d0>: Failed to establish a new connection: [Errno 111] Connection refu
sed',))

[Alex Kholkin (akholkin)]

This issue was reproduced again on the new ISO #324 after deployment regular Murano (without experimental group)

[Alexander Tivelkov (ativelkov)]

Looks like there are some issues of Horizon obtaining the service catalog from Keystone during the sign in process. This may be possibly caused by https://bugs.launchpad.net/mos/+bug/1580600 - but may be not.

[Alexander Tivelkov (ativelkov)]

Can we try to reverify this on ISO 372 or newer?

[Alexander Tivelkov (ativelkov)]

Still reproducible on ISO #390

[Alexander Tivelkov (ativelkov)]

Probably the root cause may be in the following configuration issue: https://bugs.launchpad.net/mos/+bug/1584143

[Dina Belova (dbelova)]

> Probably the root cause may be in the following configuration issue:
> https://bugs.launchpad.net/mos/+bug/1584143

Fix for this bug is under testing for mitaka.

[Dina Belova (dbelova)]

Fix for keystone-related issue landed in both master and mitaka branches.

[Alex Kholkin (akholkin)]

Still reproduce on ISO #455 9.0-mos.all.
It looks like we have the same problem as here https://bugs.launchpad.net/fuel/+bug/1589433
This could be caused by some glitches in keystone’s cache, which could be caused by this bug https://bugs.launchpad.net/fuel/+bug/1587771

[Dina Belova (dbelova)]

Nope, the https://bugs.launchpad.net/fuel/+bug/1587771 is about two issues in fact - one regarding the cache, the second one regarding https://bugs.launchpad.net/fuel/+bug/1589433 (separated to one more bug). Moreover, endpoints are not cached. So cache cannot influence at all.

[Alexander Tivelkov (ativelkov)]

So, it seems like we have to postpone the fix of this bug and move to for the 9.0-updates.

There is an easy workaround:

This bug occurs only if the user logs in to Horizon immediately (or soon enough, i.e. less then 10 minutes or smth) after the MOS deployment has finished. If the user waits some time before logging in to Horizon, the bug does not occur.

If the user logged into soon enough and the bug takes place it will continue happening till the end of the horizon session. So, we may suggest the users to logout from Horizon and relogin in some time if they encounter the described error.

This bug and the workaround should be added to 9.0 release-notes.

[Kirill Zaitsev (kzaitsev)]

Will be fixed in 9.2