image.find_image hangs on a far away cloud with lots of images when searching by ID

[Tom Fifield]

hello!

A cloud I use currently lists 429 images, and is about 230ms RTT to/from me.

Using openstacksdk 0.6.2, auth and some simple calls all work successfully, eg:
* conn.image.images()
* conn.compute.get_image(image_id)

Finding an image by name works too, eg
* conn.image.find_image('CloudMan & CBL 12.04')

However, using find_image with an ID eg:

image = conn.image.find_image('d57696ba-5ed2-43fe-bf78-a587829973a9')

it hangs and never recovers.

I've pulled out wireshark in an attempt to see what it's doing while it's sitting there.

Of course, it's all encrypted, but what I see is:
* Get /v1/images/d57696ba-5ed2-43fe-bf78-a587829973a9
* 83 seconds of packets back and forth between the glance API endpoint and the client
* a single packet from the keystone server
* no further traffic

After this 90 seconds or so from when the call was first run, the call is still there hanging.

[Everett Toews (everett-toews)]

Hey Tom. It's possible the code causing this bug was recently completely replaced when the SDK switched to using keystoneauth.

Can you please try this again with version 0.7.1 and see if the situation has improved?

[Tom Fifield (fifieldt)]

Yessir - standby!

[Tom Fifield (fifieldt)]

Just tried with version 0.7.3, same situation (lasts a bit longer than 83 seconds though).