Icehouse Glance starts but does not listen - doesn't bind to configured port 9292

[Disaster]

Ubuntu 14.04.1 LTS, all configured as described in the official installation guide up to "Install the Image service" section.

service glance-api start
and
service glance-registry start

report NO errors.
Relevant processes are confirmed running.

But neither is listening to port 9292, and so image-create in next section of the Installation Guide ("Verify the Identity Service installation") obviously fails with

[Errno 111] Connection refused

Meanwhile, in /var/log/glance/api.log, a block of traces repeats endlessly.
The block contains plenty of warnings but one Error and one Critical worth attention

2014-09-24 08:35:42.527 21646 ERROR glance.store.sheepdog [-] Error in store configuration: [Errno 2] No such file or directory
2014-09-24 08:35:42.534 21646 WARNING glance.store.base [-] Failed to configure store correctly: Store sheepdog could not be configured correctly. Reason: Error in store configuration: [Errno 2] No such file or directory Disabling add method.
2014-09-24 08:35:42.537 21646 WARNING glance.store [-] Deprecated: glance.store.sheepdog.Store not found in `known_store`. Stores need to be explicitly enabled in the configuration file.
2014-09-24 08:35:42.552 21646 WARNING glance.store [-] Deprecated: glance.store.rbd.Store not found in `known_store`. Stores need to be explicitly enabled in the configuration file.
2014-09-24 08:35:42.561 21646 WARNING glance.store.base [-] Failed to configure store correctly: Store gridfs could not be configured correctly. Reason: Missing dependencies: pymongo Disabling add method.
2014-09-24 08:35:42.564 21646 WARNING glance.store [-] Deprecated: glance.store.gridfs.Store not found in `known_store`. Stores need to be explicitly enabled in the configuration file.
2014-09-24 08:35:42.998 21646 WARNING glance.store.base [-] Failed to configure store correctly: Store cinder could not be configured correctly. Reason: Cinder storage requires a context. Disabling add method.
2014-09-24 08:35:43.002 21646 WARNING glance.store [-] Deprecated: glance.store.cinder.Store not found in `known_store`. Stores need to be explicitly enabled in the configuration file.
2014-09-24 08:35:43.036 21646 WARNING glance.store [-] Deprecated: glance.store.swift.Store not found in `known_store`. Stores need to be explicitly enabled in the configuration file.
2014-09-24 08:35:43.152 21646 WARNING glance.store [-] Deprecated: glance.store.vmware_datastore.Store not found in `known_store`. Stores need to be explicitly enabled in the configuration file.
2014-09-24 08:35:43.164 21646 WARNING glance.store [-] Deprecated: glance.store.s3.Store not found in `known_store`. Stores need to be explicitly enabled in the configuration file.
2014-09-24 08:35:44.287 21646 WARNING keystoneclient.middleware.auth_token [-] This middleware module is deprecated as of v0.10.0 in favor of keystonemiddleware.auth_token - please update your WSGI pipeline to reference the new middleware package.
2014-09-24 08:35:44.291 21646 CRITICAL glance [-] AttributeError: 'StrOpt' object has no attribute 'type'

The CRITICAL error is raised from deep within /usr/local/lib/python2.7/dist-packages/keystoneclient/middleware/auth_token.py
where the statement
opt_types = dict((o.dest, o.type) for o in opts)
fails. My guess is this must be a syntax error in some config file, leading to wrong/corrupt opts, which causes dict(..) call to fail.
However I edited the config files exactly as described in the guide.
No applicable/relevant comment found on the web for this error message and CRITICAL condition.

As for the ERROR, it is caused by collie not found. Installing sheepdog the error changes: now collie is found, but it now dislikes "localhost" as ip address. Which of the "localhost" strings that are spread over the Glance-related config files ? I didn't change them, and this guide is not requiring to change localhost to another string, so I was assuming these default config files were ok.

Most importantly, as per the official installation guide, following the instructions was supposed to result in setting up a plain file-based storage. Is sheepdog involved in this storage model, or is it activated by mistake? (again, I didn't change any sheepdog-related option in the Glance config files).

[Zhi Yan Liu (lzy-dev)]

First, you can just ignore that Sheepdog error safely due to you did configure sheepdog/collie correctly in your deployment env. IFY, in future, I'd like to leverage the native lib instead of cli way, when sheepdog python lib get ready [0].

Second, it seems this issue is caused you used a old version oslo.config (<1.3.0) with a old version python-keystoneclient which didn't include this fix [1]. So, IMO on the moment glance-api is already crashed there, pls double-check.

[0] https://pypi.python.org/pypi/Sheepdog
[1] https://github.com/openstack/python-keystoneclient/commit/dee8bc62d641f633342cfdc37a246916a40b2f33

[Erno Kuvaja (jokke)]

No response for over month, closing as invalid.