[bug] push: bare --force-with-lease ignored on remote groups
- From
- Patrick Peroša <pa3ck.perosa@gmail.com>
- Date
- Nov 3, 2026, 14:56 UTC
- Message-ID
- <CAByu3etmpxF+zNqDQ0VJ5DZs_qxdnvQxxmeaMcGRGf36+8Oqww@mail.gmail.com>
What did you do before the bug happened? (Steps to reproduce your issue)
Pushed a non-fast-forward update to a remote group using the bare --force-with-lease option (no <refname>[:<expect>] argument).
Minimal reproduction with two file-based remotes grouped as "both":
git init --bare RemoteA.git
git init --bare RemoteB.git
git init -b main work && cd work
git remote add RemoteA ../RemoteA.git
git remote add RemoteB ../RemoteB.git
git config remotes.both "RemoteA RemoteB"
echo 1 >f && git add f && git commit -m one
echo 2 >>f && git commit -am two
git push RemoteA main && git push RemoteB main
git fetch --all
git reset --hard HEAD~1
echo 3 >>f && git commit -am threeGIT_TRACE=1 git push --force-with-lease --dry-run both main
What did you expect to happen? (Expected behavior)
The --force-with-lease non fast forward push to happen to both remotes.
+ d30bde7...c21af72 main -> main (forced update)The same as would happen if the pushes were done individually
What happened instead? (Actual behavior)
Both pushes are rejected as non-fast-forward and the command exits 1:
! [rejected] main -> main (non-fast-forward)
What's different between what you expected and what actually happened?
Both pushes failed, seemingly not receiving the --force-with-lease flag
Anything else you want to add:
GIT_TRACE showed that a bare --force-with-lease does not forward the flag to the child processes that are spawned to execute the push on each remote.
12:10:42.376196 git.c:506 trace: built-in: git push --force-with-lease --dry-run both main 12:10:42.376772 run-command.c:673 trace: run_command: git push --dry-run RemoteA main Pushing to RemoteA 12:10:42.377701 run-command.c:765 trace: start_command: /root/git-builds/next/libexec/git-core/git push --dry-run RemoteA main 12:10:42.380551 git.c:506 trace: built-in: git push --dry-run RemoteA main
The issue does not happen when an explicit <refname> is specified
12:10:42.420633 git.c:506 trace: built-in: git push --force-with-lease=main --dry-run both main 12:10:42.421239 run-command.c:673 trace: run_command: git push --dry-run --force-with-lease=main RemoteA main Pushing to RemoteA 12:10:42.421423 run-command.c:765 trace: start_command: /root/git-builds/next/libexec/git-core/git push --dry-run --force-with-lease=main RemoteA main
The issue also does not impact the normal --force flag
12:10:42.437825 git.c:506 trace: built-in: git push --force --dry-run both main 12:10:42.438310 run-command.c:673 trace: run_command: git push --force --dry-run RemoteA main Pushing to RemoteA 12:10:42.438508 run-command.c:765 trace: start_command: /root/git-builds/next/libexec/git-core/git push --force --dry-run RemoteA main 12:10:42.439223 git.c:506 trace: built-in: git push --force --dry-run RemoteA main
Reproduced on: - git 2.56.0.windows.4 (Git for Windows, where I first spotted the bug) - git 2.56.0 (built from source on Linux/WSL2) - git 2.56.0.341.gebdc087dc9 (master, built from source on Linux/WSL2) - git 2.56.0.506.g506a7eabc5 (next, built from source on Linux/WSL2)
--dry-run is used above only to make the reproduction repeatable; the same rejection happens without it.
[System Info] git version: git version 2.56.0.506.g506a7eabc5 cpu: x86_64 built from commit: 506a7eabc523d0811286b66193419658da95cda5 sizeof-long: 8 sizeof-size_t: 8 shell-path: /bin/sh rust: enabled feature: fsmonitor--daemon libcurl: 8.18.0 OpenSSL: OpenSSL 3.5.5 27 Jan 2026 zlib: 1.3.1 SHA-1: SHA1_DC SHA-256: SHA256_BLK default-ref-storage-format: files default-hash: sha1 uname: Linux 6.6.87.2-microsoft-standard-WSL2 #1 SMP PREEMPT_DYNAMIC Thu Jun 5 18:30:46 UTC 2025 x86_64 compiler info: gnuc: 15.2 libc info: glibc: 2.43 $SHELL (typically, interactive shell): /bin/bash
[Enabled Hooks]