> Excellent. That looks like a fine fall back. > > Not having looked at the C code, > have you considered if another dir/files are already setup with inotify > watches, > and if files are deleted but still open? > Other comments below... >
I changed the behaviour of tail to exit inotify mode if at least one of the watched directories is deleted. If I understand you correctly, there should not be problem with files that are deleted while being open. The files are tailed again in the polling mode, after the path leading to them is recreated(tail -F uses following by name not by fd). I also thought about entering inotify mode again after all watched directories(and the files) are found by polling mode. What I made so far, contained too many deleted and recreated watches due to exiting inotify(deleting watches to all inotify watch descriptors) mode immediately after at least one watched directory is removed. I have not found any discussion about such solution, therefore I am unsure about the safety of the solution so far. > Why this sleep? > > > +timeout 60 tail --follow=name --retry dir/file &>out & pid=$! > > +sleep 1 > > We can't depend on absolute sleeps like this. See: > http://www.pixelbeat.org/docs/coreutils-testing.html#backoff > or the existing tail-2/ tests > I apologize, thanks for the tip. See V2 patch attached with correct usage of sleep by *retry_delay_ ().* *Sebastian.*
From 0930c2dbdb50496f5bfe2b370808ec1c3fcaabda Mon Sep 17 00:00:00 2001 From: Sebastian Kisela <[email protected]> Date: Wed, 5 Apr 2017 09:40:41 +0200 Subject: [PATCH] tail: 'tail -F dir/file' reverts to polling mode if 'dir' is removed * src/tail.c (tail_forever_inotify): Add the IN_DELETE_SELF flag when creating watch for the parent directory. After the parent directory is removed, an event is caught and then we switch from inotify to polling mode. Till now, inotify has always frozen because it waited for an event from a watched dir, which has been already deleted and was not added again. * tests/tail-2/inotify-dir-recreate.sh: Add a test case. * tests/local.mk: Reference the new test. * NEWS: Mention the bug fix. Reported at https://bugzilla.redhat.com/1283760 --- NEWS | 2 + src/tail.c | 21 ++++++++- tests/local.mk | 1 + tests/tail-2/inotify-dir-recreate.sh | 84 ++++++++++++++++++++++++++++++++++++ 4 files changed, 107 insertions(+), 1 deletion(-) create mode 100755 tests/tail-2/inotify-dir-recreate.sh diff --git a/NEWS b/NEWS index e2f298f..f34d667 100644 --- a/NEWS +++ b/NEWS @@ -74,6 +74,8 @@ GNU coreutils NEWS -*- outline -*- 'cp -fl A B' no longer remove B before creating the new link. That is, there is no longer a brief moment when B does not exist. + tail -F 'dir/file' reverts to polling mode in case 'dir' is removed. + ** New features expand and unexpand now support specifying a tab size to use diff --git a/src/tail.c b/src/tail.c index d1552d4..e35ee1e 100644 --- a/src/tail.c +++ b/src/tail.c @@ -1457,7 +1457,8 @@ tail_forever_inotify (int wd, struct File_spec *f, size_t n_files, In that case the same watch descriptor is returned. */ f[i].parent_wd = inotify_add_watch (wd, dirlen ? f[i].name : ".", (IN_CREATE | IN_DELETE - | IN_MOVED_TO | IN_ATTRIB)); + | IN_MOVED_TO | IN_ATTRIB + | IN_DELETE_SELF)); f[i].name[dirlen] = prev; @@ -1628,6 +1629,24 @@ tail_forever_inotify (int wd, struct File_spec *f, size_t n_files, ev = void_ev; evbuf_off += sizeof (*ev) + ev->len; + /* If a directory is deleted, IN_DELETE_SELF is emmited + with ev->name of length 0. + We need to catch it, otherwise it would wait forever, + as wd for directory becomes inactive. Revert to polling now. */ + if ((ev->mask & IN_DELETE_SELF) && !ev->len) + { + for (i = 0; i < n_files; i++) + { + if (ev->wd == f[i].parent_wd) + { + hash_free (wd_to_name); + error (0, 0, + _("directory containing watched file was removed")); + return true; + } + } + } + if (ev->len) /* event on ev->name in watched directory. */ { size_t j; diff --git a/tests/local.mk b/tests/local.mk index 9f1a853..bcf70c2 100644 --- a/tests/local.mk +++ b/tests/local.mk @@ -176,6 +176,7 @@ all_tests = \ tests/tail-2/descriptor-vs-rename.sh \ tests/tail-2/inotify-rotate.sh \ tests/tail-2/inotify-rotate-resources.sh \ + tests/tail-2/inotify-dir-recreate.sh \ tests/chmod/no-x.sh \ tests/chgrp/basic.sh \ tests/rm/dangling-symlink.sh \ diff --git a/tests/tail-2/inotify-dir-recreate.sh b/tests/tail-2/inotify-dir-recreate.sh new file mode 100755 index 0000000..083e4eb --- /dev/null +++ b/tests/tail-2/inotify-dir-recreate.sh @@ -0,0 +1,84 @@ +#!/bin/sh +# Makes sure, inotify will switch to polling mode if directory +# of the watched file was removed and recreated. +# (...instead of getting stuck forever) + +# Copyright (C) 2006-2017 Free Software Foundation, Inc. + +# This program is free software: you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation, either version 3 of the License, or +# (at your option) any later version. + +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. + +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. + +. "${srcdir=.}/tests/init.sh"; path_prepend_ ./src +print_ver_ tail + + +# Terminate any background tail process +cleanup_() { kill $pid 2>/dev/null && wait $pid; rm -r dir; } + +cleanup_fail_ () +{ + warn_ $1 + cleanup_ + fail=1 +} + +# $check_re - string to be found +# $check_f - file to be searched +check_tail_output_ () +{ + local delay="$1" + grep $check_re $check_f > /dev/null || + { sleep $delay ; return 1; } +} + +grep_timeout_ () +{ + check_re="$1" + check_f="$2" + retry_delay_ check_tail_output_ .1 5 +} + +# Prepare the file to be watched +mkdir dir && touch dir/file && echo "inotify" > dir/file +grep_timeout_ "inotify" "dir/file" || +{ cleanup_fail_ 'file was not created'; } + +#tail must print content of the file to stdout, verify +timeout 60 tail -F dir/file &>out & pid=$! +grep_timeout_ "inotify" "out" || +{ cleanup_fail_ 'file to be tailed does not exist'; } + +# Remove the directory, should get the massage about the deletion +rm -r dir +grep_timeout_ "polling" "out" || +{ cleanup_fail_ 'tail did not switch to polling mode'; } + +# Recreate the dir, must get a message about recreation +mkdir dir && touch dir/file +grep_timeout_ "appeared" "out" || +{ cleanup_fail_ 'previously removed file did not appear'; } + +cleanup_ + +# Expected result for the whole process +cat <<\EOF > exp +inotify +tail: 'dir/file' has become inaccessible: No such file or directory +tail: directory containing watched file was removed +tail: inotify cannot be used, reverting to polling: No such file or directory +tail: 'dir/file' has appeared; following new file +EOF + +compare exp out || fail=1 + +Exit $fail -- 2.9.3
