Fixes salt.utils.EtcdClient.write so it distinguishes between the different python-etcd logic for files/directories

[arizvisa]

What does this PR do?

Fixes some oversights in the salt.utils.etcd_util module when creating directories vs files.

What issues does this PR fix or reference?

Closes issue #51537

Previous Behavior

When creating a directory an error message is logged saying the path is not a file.

New Behavior

Fixes the prior mentioned behaviour.

Tests written?

I'm tired.

Commits signed with GPG?

No.

[arizvisa]

This PR was only written for the file-directory duality in the salt.utils.etcd_util module and does not affect error messages that are emitted when an empty directory entry is attempted to be read.

This is done by the salt.states.etcd_mod in order to determine the original value to compare to. This function will pass the recurse=True option to salt.utils.etcd_util.read() in an attempt to read the contents of the directory. Due to the directory being empty, the etcd client will raise an EtcdKeyNotFound exception, which will then get logged before the .read() function properly returns None. This is a red herring in that no error has happened, there are just no "files" in the directory.

Nonetheless, this only fixes the salt.utils.EtcdClient.write() method since other things depend on it.

[dwoz]

@arizvisa are you able to create a test that verifies this functionality? I think it will prevent us from having a regression in the future.

[arizvisa]

I'll see what I can do in the future, it'll take some time though as my primary job isn't related to development or even close to this sorta devops stuff so it's hard to find business cause for this sorta thing.

Honestly I expected using saltstack was to be a quick effort, but I ended up spending a lot more time on this due to the Etcd support apparently being underused (Such as #51345 which you were a part of, then PRs #51363 and #51346). Still a fan of the whole design of salt, just wasn't expecting to do so much work for my one very-specific configuration.

[arizvisa]

Oh also, I think I mentioned the issue with the salt.state.etcd_mod.directory state calling the .get() method on a directory that doesn't exist and it not catching the expected error. That will probably need to get caught at some point and demoted to a log.trace as well.

I didn't create the issue yet because I don't have too much time to fix and test it. So just a heads up if someone else gets to writing tests for this PR before I do, they'll still see an error in the logs when using etcd.directory on a path that needs to be created.

[dwoz]

@arizvisa One of the existing Etcd unit tests is failing, This needs to be fixed before we can merge.

[arizvisa]

force-push due to a rebase against develop. added unit tests for both keys and directories. we'll see how it turns out..

[arizvisa]

Grr...first force-push was against develop instead of origin/develop. This next push will be against the correct branch...

[arizvisa]

Um...am I reading this error right from py2-windows-2016? It seems to have terminated before it even ran my tests?

>>>>>> ------Exception-------
>>>>>> Class: Kitchen::ActionFailed
>>>>>> Message: 1 actions failed.
>>>>>>     Failed to complete #verify action: [[WSMAN ERROR CODE: 2152992672]: <f:WSManFault Code='2152992672' Machine='10.200.20.199' xmlns:f='http://schemas.microsoft.com/wbem/wsman/1/wsmanfault'><f:Message><f:ProviderFault path='C:\Windows\system32\pwrshplugin.dll' provider='microsoft.powershell'>Exception of type &apos;System.OutOfMemoryException&apos; was thrown.</f:ProviderFault></f:Message></f:WSManFault>] on py2-windows-2016

[dwoz]

Um...am I reading this error right from py2-windows-2016? It seems to have terminated before it even ran my tests?

>>>>>> ------Exception-------
>>>>>> Class: Kitchen::ActionFailed
>>>>>> Message: 1 actions failed.
>>>>>>     Failed to complete #verify action: [[WSMAN ERROR CODE: 2152992672]: <f:WSManFault Code='2152992672' Machine='10.200.20.199' xmlns:f='http://schemas.microsoft.com/wbem/wsman/1/wsmanfault'><f:Message><f:ProviderFault path='C:\Windows\system32\pwrshplugin.dll' provider='microsoft.powershell'>Exception of type &apos;System.OutOfMemoryException&apos; was thrown.</f:ProviderFault></f:Message></f:WSManFault>] on py2-windows-2016

Yes, we still have some issues in the test runner. Having said that, it looks like the next test run passed. :) Merging.

[arizvisa]

Thx.