Description of problem: 1. find is being used on NFS file systems. 2. two processes e.g. 'find DIR -name PATTERN -ignore_readdir_race -exec rm {} \;' look for files to delete in the same directory. 3. one archive process works as expected. 4. the other archive process running 'find' reports error message 'find: '/path/to/file': No such file or directory' Note, only shows up on NFS file systems not local file systems. Version-Release number of selected component (if applicable): - findutils-4.6.0-20.el8.x86_64 How reproducible: Sometimes Steps to Reproduce: 1. Create two find processes deleting files as above. 2. Run many times, sometimes find reports error 'find: '/path/to/file': No such file or directory' Actual results: 'find: '/path/to/file': No such file or directory' Expected results: With option -ignore_readdir_race find should ignore missing files when trying to stat() them. Additional info: - high business impact for customer. Blocking migration. - strace performed confirming behaviour and attached to case. See extract below. ~~~ 231057 18:30:01.360720 getdents64(4</infadata/Backup/FIND_POC/Target>, [{d_ino=154801, d_off=11878423, ... d_name="MC_EDH_SIDE_CAR_GCP_POST_VBUP20230730-02:18:29_SmLYdODCg7.log"}] ... 231057 18:30:27.998095 newfstatat(5</infadata/Backup/FIND_POC/Target>, "MC_EDH_SIDE_CAR_GCP_POST_VBUP20230730-02:18:29_SmLYdODCg7.log", 0x55dc79775fe8, AT_SYMLINK_NOFOLLOW) = -1 ENOENT (No such file or directory) <0.000327> 231057 18:30:27.999508 newfstatat(5</infadata/Backup/FIND_POC/Target>, "MC_EDH_SIDE_CAR_GCP_POST_VBUP20230730-02:18:29_SmLYdODCg7.log", 0x7ffe36e02980, AT_SYMLINK_NOFOLLOW) = -1 ENOENT (No such file or directory) <0.000343> 231057 18:30:28.000927 write(2</infadata/Backup/FIND_POC/Log/find_and_remove_GLT0.log>, "find: ", 6) = 6 <0.000416> 231057 18:30:28.002800 write(2</infadata/Backup/FIND_POC/Log/find_and_remove_GLT0.log>, "\342\200\230/infadata/Backup/FIND_POC/Target/MC_EDH_SIDE_CAR_GCP_POST_VBUP20230730-02:18:29_SmLYdODCg7.log\342\200\231", 100) = 100 <0.000506> 231057 18:30:28.004705 write(2</infadata/Backup/FIND_POC/Log/find_and_remove_GLT0.log>, ": No such file or directory", 27) = 27 <0.000412> [snip] 231057 18:30:30.384816 +++ exited with 1 +++ ~~~
Hello! Thank you for the report! I was able to reproduce this issue on my machine. This problem was already reported to upstream few years back (https://savannah.gnu.org/bugs/?45930) but the fix was not committed upstream, yet. So it is possible to encounter it on any release of RHEL or Fedora (even RHEL 7), it just happens that NFS on RHEL 8 has "better" timings to trigger it. Fortunately, the fix seems trivial. I'll update you ASAP when I have additional information to share. Regards, Lukas
C8S MR: https://gitlab.com/redhat/centos-stream/rpms/findutils/-/merge_requests/3