Fix list_nodes in the Azure arm driver

[jan-mue]

Fix the list_nodes function in the Azure ARM driver

Description

The Azure API returns an attribute nextLink for pagination that is currently not used in the Azure ARM driver.
Documentation: https://learn.microsoft.com/en-us/rest/api/compute/virtual-machines/list-all?tabs=HTTP

This PR fixes the issue where not all VMs are returned for subscriptions with a lot of VMs (#1824)

Status

ready for review

Checklist (tick everything that applies)

• Code linting (required, can be done after the PR checks)
• Documentation
• Tests
• ICLA (required for bigger changes)

[Kami]

Thanks for the contribution.

It would be great if you could also add a unit test for this change.

[jan-mue]

@Kami sorry for the long delay, but I've added some unit tests now. Could you please have another look at the PR?

[Kami]

@jan-mue Thanks for adding a test case. I will review it shortly and if everything looks fine, merge changes into trunk.

[Kami]

I think this code is far for now, but to be on the safe side and avoid possible infinite loop, I think it would be good to add additional guard here - e.g. bail out if we are still stuck in this loop for more than X seconds.

[Kami]

@jan-mue Can you please also sign an ICLA when you get a chance (https://libcloud.readthedocs.io/en/latest/development.html#contributing-bigger-changes)? Thanks.

[Kami]

I pushed a change so we don't end up in an infinite loop in case server returns a bad response, there is a bug in the code or similar - c49ecd0.

I picked that 60 seconds limit arbitrarily, but if needed or wanted, it can be changed (e.g. increased) - @jan-mue Can you please advise if 60 seconds is enough or the limit should be higher?

I imagine we may want a higher limit since depending on the method arguments, each _to_node() call will result in additional HTTP request(s).

For now, in case deadline is reached we simply return, but we may also want to throw or at least log a warning in such scenario.

[Kami]

EDIT: I went ahead and bumped that limit to 300 seconds for now. This way I can wrap up and merge this PR.

And then based on your future feedback @jan-mue, we can adjust this value.

[codecov-commenter]

Codecov Report

Merging #1850 (863b88d) into trunk (d163f9b) will increase coverage by 0.01%.
Report is 30 commits behind head on trunk.
The diff coverage is 89.39%.

@@            Coverage Diff             @@
##            trunk    #1850      +/-   ##
==========================================
+ Coverage   83.11%   83.12%   +0.01%     
==========================================
  Files         353      353              
  Lines       81269    81317      +48     
  Branches     8587     8592       +5     
==========================================
+ Hits        67541    67588      +47     
- Misses      10922    10924       +2     
+ Partials     2806     2805       -1     

Files Changed	Coverage Δ
libcloud/backup/drivers/dimensiondata.py	94.94% <ø> (ø)
libcloud/common/aliyun.py	74.44% <ø> (ø)
libcloud/common/azure.py	86.30% <ø> (ø)
libcloud/common/azure_arm.py	73.02% <ø> (ø)
libcloud/common/durabledns.py	99.25% <ø> (ø)
libcloud/common/google.py	71.35% <ø> (ø)
libcloud/common/gridscale.py	83.93% <ø> (ø)
libcloud/common/kubernetes.py	80.81% <ø> (ø)
libcloud/common/nttcis.py	73.59% <ø> (ø)
libcloud/common/vultr.py	84.93% <ø> (ø)
... and 69 more