erepo: ignores ssh-agent keys in windows git-bash #7702
Assignees
Labels
bug
Did we break something?
git
Related to git and git backends
P: windows
Related to the Platform: Windows
regression
Ohh, we broke something :-(
Comments
Following up: the data pulled in |
daavoo
added
bug
|
@piotlinski Thanks for the report. Could you share a little more traceback detail from |
|
@daavoo Sure, here is the output of **Click to show spoiler**btw. calling |
This comment was marked as resolved.
This comment was marked as resolved.
|
This is not the same issue. #7670 is due to git credential storage being unsupported. This looks like a problem with asyncssh support for ssh-agent on windows. @piotlinski what shell are you using in windows? edit: I see this is in github actions, I'm not sure how GHA translates unix style paths in actions (like |
|
@pmrowla I did try with a dedicated action for setting up ssh-agent: https://github.com/webfactory/ssh-agent (I supposed it should handle every OS available @ GitHub Actions properly), but the error occurred as well. With previous DVC versions we did not have any issues with the approach presented above. |
Meant that because 2.10 is the version that introduced the issue appears to be related to adopting |
|
@daavoo Yes, adopting |
|
Hi @mdekstrand . @dtrifiro is working on adding support for credential helpers in dulwich, see #7307 (comment) :) |
|
@daavoo Yep, seeing that issue is how I knew the HTTPS side was a dulwich problem. At least until that PR lands in a published release, could |
The only way (disclaimer: it's hacky 😅 ) I can think of achieving this without requiring a new release (either dvc, scmrepo or dulwich) would be for you to edit your local I think it might be better to wait for the release or downgrade dvc
I think having the ability to select a git backend would be a valid feature request for scmrepo. However, I believe that the current direction in DVC is to focus on using/supporting only |
This comment was marked as off-topic.
This comment was marked as off-topic.
|
@dtrifiro Any thoughts? |
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
|
@dtrifiro why closing the issue? Unfortunately the problem is not resolved - on latest version (2.27.2) the problem still exists in GitHub Actions, as described in the original posts. I attach the logs below: click to expand logs2022-09-27 10:24:38,151 DEBUG: Creating external repo git@github.com:***/project-benchmarks.git@270d169b792ed8a8e02ef859dd80ae47b7febd18
2022-09-27 10:24:38,151 DEBUG: erepo: git clone 'git@github.com:***/project-benchmarks.git' to a temporary dir
2022-09-27 10:24:40,356 ERROR: failed to pull data from the cloud - Failed to clone repo 'git@github.com:***/project-benchmarks.git' to 'C:\Users\RUNNER~1\AppData\Local\Temp\tmpltcso3mudvc-clone'
------------------------------------------------------------
Traceback (most recent call last):
File "D:\a\project\project\.venv\lib\site-packages\scmrepo\git\backend\dulwich\__init__.py", line 200, in clone
repo = clone_from()
File "D:\a\project\project\.venv\lib\site-packages\dulwich\porcelain.py", line 538, in clone
return client.clone(
File "D:\a\project\project\.venv\lib\site-packages\dulwich\client.py", line 760, in clone
result = self.fetch(path, target, progress=progress, depth=depth)
File "D:\a\project\project\.venv\lib\site-packages\dulwich\client.py", line 837, in fetch
result = self.fetch_pack(
File "D:\a\project\project\.venv\lib\site-packages\dulwich\client.py", line 1146, in fetch_pack
proto, can_read, stderr = self._connect(b"upload-pack", path)
File "D:\a\project\project\.venv\lib\site-packages\dulwich\client.py", line 1792, in _connect
con = self.ssh_vendor.run_command(
File "D:\a\project\project\.venv\lib\site-packages\fsspec\asyn.py", line 111, in wrapper
return sync(self.loop, func, *args, **kwargs)
File "D:\a\project\project\.venv\lib\site-packages\fsspec\asyn.py", line 96, in sync
raise return_result
File "D:\a\project\project\.venv\lib\site-packages\fsspec\asyn.py", line 53, in _runner
result[0] = await coro
File "D:\a\project\project\.venv\lib\site-packages\scmrepo\git\backend\dulwich\asyncssh_vendor.py", line 163, in _run_command
conn = await asyncssh.connect(
File "D:\a\project\project\.venv\lib\site-packages\asyncssh\connection.py", line 7834, in connect
return await asyncio.wait_for(
File "C:\hostedtoolcache\windows\Python\3.10.4\x64\lib\asyncio\tasks.py", line 408, in wait_for
return await fut
File "D:\a\project\project\.venv\lib\site-packages\asyncssh\connection.py", line 447, in _connect
await options.waiter
asyncssh.misc.PermissionDenied: Permission denied
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
2022-09-27 10:24:40,372 DEBUG: Analytics is disabled.
File "D:\a\project\project\.venv\lib\site-packages\dvc\scm.py", line 145, in clone
git = Git.clone(url, to_path, progress=pbar.update_git, **kwargs)
File "D:\a\project\project\.venv\lib\site-packages\scmrepo\git\__init__.py", line 143, in clone
backend.clone(url, to_path, **kwargs)
File "D:\a\project\project\.venv\lib\site-packages\scmrepo\git\backend\dulwich\__init__.py", line 203, in clone
raise CloneError(url, to_path) from exc
scmrepo.exceptions.CloneError: Failed to clone repo 'git@github.com:***/project-benchmarks.git' to 'C:\Users\RUNNER~1\AppData\Local\Temp\tmpltcso3mudvc-clone'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "D:\a\project\project\.venv\lib\site-packages\dvc\commands\data_sync.py", line 31, in run
stats = self.repo.pull(
File "D:\a\project\project\.venv\lib\site-packages\dvc\repo\__init__.py", line 49, in wrapper
return f(repo, *args, **kwargs)
File "D:\a\project\project\.venv\lib\site-packages\dvc\repo\pull.py", line 34, in pull
processed_files_count = self.fetch(
File "D:\a\project\project\.venv\lib\site-packages\dvc\repo\__init__.py", line 49, in wrapper
return f(repo, *args, **kwargs)
File "D:\a\project\project\.venv\lib\site-packages\dvc\repo\fetch.py", line 82, in fetch
used = self.used_objs(
File "D:\a\project\project\.venv\lib\site-packages\dvc\repo\__init__.py", line 430, in used_objs
for odb, objs in self.index.used_objs(
File "D:\a\project\project\.venv\lib\site-packages\dvc\repo\index.py", line 233, in used_objs
for odb, objs in stage.get_used_objs(
File "D:\a\project\project\.venv\lib\site-packages\dvc\stage\__init__.py", line 689, in get_used_objs
for odb, objs in out.get_used_objs(*args, **kwargs).items():
File "D:\a\project\project\.venv\lib\site-packages\dvc\output.py", line 1034, in get_used_objs
return self.get_used_external(**kwargs)
File "D:\a\project\project\.venv\lib\site-packages\dvc\output.py", line 1089, in get_used_external
return dep.get_used_objs(**kwargs)
File "D:\a\project\project\.venv\lib\site-packages\dvc\dependency\repo.py", line 97, in get_used_objs
used, _ = self._get_used_and_obj(**kwargs)
File "D:\a\project\project\.venv\lib\site-packages\dvc\dependency\repo.py", line 111, in _get_used_and_obj
with self._make_repo(locked=locked, cache_dir=local_odb.path) as repo:
File "C:\hostedtoolcache\windows\Python\3.10.4\x64\lib\contextlib.py", line 135, in __enter__
return next(self.gen)
File "D:\a\project\project\.venv\lib\site-packages\dvc\external_repo.py", line 39, in external_repo
path = _cached_clone(url, rev, for_write=for_write)
File "D:\a\project\project\.venv\lib\site-packages\dvc\external_repo.py", line 169, in _cached_clone
clone_path, shallow = _clone_default_branch(url, rev, for_write=for_write)
File "D:\a\project\project\.venv\lib\site-packages\funcy\decorators.py", line 45, in wrapper
return deco(call, *dargs, **dkwargs)
File "D:\a\project\project\.venv\lib\site-packages\funcy\flow.py", line 274, in wrap_with
return call()
File "D:\a\project\project\.venv\lib\site-packages\funcy\decorators.py", line 66, in __call__
return self._func(*self._args, **self._kwargs)
File "D:\a\project\project\.venv\lib\site-packages\dvc\external_repo.py", line 239, in _clone_default_branch
git = clone(url, clone_path)
File "D:\a\project\project\.venv\lib\site-packages\dvc\scm.py", line 150, in clone
raise CloneError(str(exc))
dvc.scm.CloneError: Failed to clone repo 'git@github.com:***/project-benchmarks.git' to 'C:\Users\RUNNER~1\AppData\Local\Temp\tmpltcso3mudvc-clone'btw. version 2.9.5 works perfectly fine in the same setup. |
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
|
@piotlinski The github actions snippet you posted does not seem to be working, this is the error I'm getting on gha when running your code: Run ssh-agent -a $SSH_AUTH_SOCK > /dev/null # breaks here
ssh-agent -a $SSH_AUTH_SOCK > /dev/null # breaks here
ssh-keygen -t ed25519 -N dummy -f key_ed25519 # creating a new key
ssh-add key_ed25519 #
mkdir -p ~/.ssh/
ssh-keyscan -H github.com > ~/.ssh/known_hosts
shell: C:\Program Files\PowerShell\7\pwsh.EXE -command ". '{0}'"
env:
SSH_AUTH_SOCK: /tmp/ssh_agent.sock
Out-File: D:\a\_temp\1dfe347a-6052-440e-ac06-4369f0d0a4ce.ps1:2
Line |
2 | ssh-agent -a $SSH_AUTH_SOCK > /dev/null # breaks here
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| Could not find a part of the path 'D:\dev\null'.which makes sense to me, since it's trying to access |
Sorry, forgot to include one part: You need to be running this in bash and not powershell |
dberenbaum
removed
the
p1-important
This was referenced Jan 30, 2023
|
It looks like this is specific to git-bash/mingw openssh (which is what is used when you specify As far as I can tell, the problem is that the asyncssh windows agent implementation expects that it will be used with a windows pipe (i.e. UNC path I'm guessing that this would actually work out of the box if you were using a mingw python build, but the problem here happens because we end up mixing a native windows python build inside a mingw shell environment. The fix is probably for us to just check for mingw environment vars when we are in windows and then fallback to the subprocess |
pmrowla
changed the title
pmrowla
removed
research
upstream
|
@pmrowla Great research! 🔥 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Bug Report
pull: ignores ssh-agent keys
Description
DVC 2.10.0 stops using ssh-agent (passed via
SSH_AUTH_SOCKvariable) for git clone operation resulting inPermission deniederror.Reproduce
Expected
Artefacts are downloaded as in <=2.9.5 version
Environment information
GitHub Actions
The text was updated successfully, but these errors were encountered: