On 19/04/18 07:43, Jewsco Pius Jacquez wrote: > Pádraig, thanks for your response. > > In my reproducer environment, it doesn't totally update. RHEL7 Samba server, > RHEL6 Samba client. > > Unless I provided the extra interval parameter.
You said you used: tail --follow=name --max-unchanged-stats=1 --sleep-interval=1 The --sleep-interval=1 having a difference is confusing as that already defaults to one second. Also --max-unchanged-stats=1 should only make things more responsive. I would have expected only --follow=name to have a significant difference to the data that is eventually output. Anyway there may be an issue with st_size updating on your system. I.E. if stat() didn't update st_size on this remote file then a new change that protects against a related issue on glusterfs may be impacting here. One could check that by reverting the patch and rebuilding in your previous build dir like: cd coreutils-8.29 && wget 'https://github.com/coreutils/coreutils/commit/7fc7206.patch' -O- | patch -R -p1 && make One might also check this externally to tail(1) by repeatedly running the following and seeing does the number correlate with remote changes: stat -c '%s' /media/samba/test.file cheers, Pádraig
