Strip trailing slash before DFS GetProperties (fixes OneLake 403) (backport #174)#177
Merged
benfleis merged 5 commits intoJul 13, 2026
Conversation
plus other aesthetic cleanups
A trailing '/' is a directory hint, not part of the resource name, but it was passed straight into GetFileClient(parsed_url.path). OneLake's DFS endpoint rejects a GetProperties on a trailing-slash path with 403 AuthenticationFailed (the slash-less path returns 200), so any DirectoryExists/FileExists/glob on a directory fails. This breaks e.g. duckdb-delta partitioned writes, which probe the table directory. Strip a single trailing slash from the path used to build the file client. info.path is left untouched, so the directory-marker check in LoadRemoteFileInfo still works. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Member
Author
|
@djouallah this is the v1.5 variant of #174; it should be available in core_nightly for v1.5.4 in a couple hours. |
Contributor
|
The v1.5 cherry-pick dropped the call-site change — CreateHandle still passes parsed_url.path to GetFileClient, so the strip block is dead code and the 403 still repros on 334798f. Needs GetFileClient(file_path) like main. One-liner: https://github.com/duckdb/duckdb-azure/blob/v1.5-variegata/src/azure_dfs_filesystem.cpp (line ~174) |
benfleis
added a commit
to benfleis/duckdb-azure
that referenced
this pull request
Jul 14, 2026
benfleis
added a commit
that referenced
this pull request
Jul 14, 2026
Re PR #177 , noted [here](#177 (comment))
Member
Author
|
@djouallah thanks for the catch, the fix is pushed out and will be pulled in from nightly. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
(backport #174), thanks @djouallah
Problem
On the DFS filesystem,
CreateHandlepassesparsed_url.pathstraight intoGetFileClient(...), so a trailing/ends up on the wire in the subsequentGetProperties()call. A trailing slash is only a directory hint, not part of the resource name.OneLake's DFS endpoint (
*.dfs.fabric.microsoft.com) rejects aGetPropertieson a trailing-slash path with403 AuthenticationFailed/Forbidden, while the identical request without the slash returns200. BecauseDirectoryExists,FileExists, andGlob(no-wildcard) all funnel throughOpenFile→CreateHandle→LoadRemoteFileInfo→GetProperties, any directory probe on OneLake fails.This surfaces in practice as duckdb-delta partitioned writes failing:
PhysicalCopyToFileprobes the target table directory viaDirectoryExists("…/Tables/dbo/<table>/"), producing:Reproduces standalone, no write needed:
Fix
Strip a single trailing
/from the path used to build the file client.info.pathis left untouched, so the directory-marker detection inLoadRemoteFileInfo(StringUtil::EndsWith(afh.GetPath(), "/")) still works; only what is sent to the service changes. Real file keys never end in/, so reads/writes/Createare unaffected.