-
Bug
-
Resolution: Fixed
-
Medium
git lfs fetch has been failing on specific files after fresh clone.
It is failing consistently on specific files so as an experiment I {{git rm}}ed the failing files, but now it's failing on a different file. There's already an issue open with git-lfs, but this feels like an LFS server issue (the first three comments on the issue mention that they use bitbucket).
Concretely,
$ GIT_LFS_SKIP_SMUDGE=1 git clone git@bitbucket.org:<repo>.git succeeds
but
git lfs fetch fails with
$ git lfs fetch fetch: Fetching reference refs/heads/master 2019/01/30 09:27:46 Unsolicited response received on idle HTTP channel starting with "<part-of-file>"; err=<nil> Downloading LFS objects: 36% (1198/3313), 1.7 GB | 4.6 MB/s Expected OID <some-hash>, got <other-hash> after 83429376 bytes written error: failed to fetch some objects from 'https://bitbucket.org/<repo>.git/info/lfs'
Rerunning with GIT_TRACE produces
#!log $ GIT_TRACE=1 git lfs fetch 09:40:57.915993 trace git-lfs: exec: git '-c' 'filter.lfs.smudge=' '-c' 'filter.lfs.clean=' '-c' 'filter.lfs.process=' '-c' 'filter.lfs.required=false' 'rev-parse' 'HEAD' '--symbolic-full-name' 'HEAD' 09:40:57.917093 trace git-lfs: tq: running as batched queue, batch size of 100 09:40:57.917103 trace git-lfs: fetch <file-name> [<id>] 09:40:57.917124 trace git-lfs: tq: sending batch of size 1 09:40:57.917166 trace git-lfs: run_command: ssh -- git@bitbucket.org git-lfs-authenticate <repo> download 09:40:59.010400 trace git-lfs: api: batch 1 files 09:40:59.010777 trace git-lfs: HTTP: POST https://bitbucket.org/.../.../info/lfs/objects/batch 09:40:59.404101 trace git-lfs: HTTP: 200 09:40:59.404185 trace git-lfs: HTTP: {"objects": [{"oid": "<id>", "actions": {"download": {"header": {"X-Client-ID": "<hash-A>", "Authorization": "Bearer <hash-B>"}, "href": "ht 09:40:59.404287 trace git-lfs: HTTP: tps://api.media.atlassian.com/file/<hash-C>/binary"}}, "size": 83429376}]} 09:40:59.404469 trace git-lfs: tq: starting transfer adapter "basic" 09:40:59.629873 trace git-lfs: xfer: Attempting to resume download of "<id>" from byte 83429376 09:40:59.630036 trace git-lfs: HTTP: GET https://api.media.atlassian.com/file/<hash-C>/binary 09:41:00.239427 trace git-lfs: HTTP: 400 09:41:00.239605 trace git-lfs: HTTP: {"error":{"code":"InvalidArgumentError","title":"range first index must not be greater than last index","href":"https://api.media.atlassian.com#InvalidArgumentError"}} 09:41:00.239809 trace git-lfs: tq: retrying object <id>: LFS: Client error: https://api.media.atlassian.com/file/<hash-C>/binary
[BCLOUD-18064] git lfs fetch fails with InvalidArgumentError
I contacted support directly and they issues a fix which has done the trick for me.
I contacted bitbucket support and they've fixed the issue for me! Here is their response:
We have been able to reproduce this issue on our end. We have identified that this issue is triggered by slow network connections at the time of pulling down the LFS files. This leads to a timeout on our backend. I would like to inform that our team have identified the issue and have deployed a fix in regards to this.
Hey - I have been traveling for work. I will have to check this next week.