From a7356806f9f7c148d4e0a972f6a418d1ca82bcd0 Mon Sep 17 00:00:00 2001 From: Jeff Darcy Date: Tue, 8 Jul 2014 21:56:04 -0400 Subject: dht: fix rename race If two clients try to rename the same file at the same time, we sometimes end up with *no file at all* in either the old or new location. That's kind of bad. The culprit seems to be some overly aggressive cleanup code. AFAICT, based on today's study of the code, the intent of the changed section is to remove any linkfile we might have created before the actual rename. However, what we're removing might not be our extra link. If we're racing with another client that's also doing a rename, it might be the only remaining link to the user's data. The solution, which is good enough to pass this test but almost certainly still not complete, is to be more selective about when we do this unlink. Now, we only do it if we know that, at some point, we did in fact create the link without error (notably ENOENT on the source or EEXIST on the destination) ourselves. Change-Id: I8d8cce150b6f8b372c9fb813c90be58d69f8eb7b BUG: 1139988 Signed-off-by: Jeff Darcy Reviewed-on: http://review.gluster.org/8269 Tested-by: Gluster Build System Reviewed-by: Vijay Bellur Reviewed-on: http://review.gluster.org/8672 Reviewed-by: Raghavendra G Reviewed-by: Kaleb KEITHLEY --- tests/bugs/bug-1117851.t | 102 +++++++++++++++++++++++++++++++++++ xlators/cluster/dht/src/dht-common.h | 1 + xlators/cluster/dht/src/dht-rename.c | 18 +++++-- 3 files changed, 117 insertions(+), 4 deletions(-) create mode 100755 tests/bugs/bug-1117851.t diff --git a/tests/bugs/bug-1117851.t b/tests/bugs/bug-1117851.t new file mode 100755 index 00000000000..194d39c61eb --- /dev/null +++ b/tests/bugs/bug-1117851.t @@ -0,0 +1,102 @@ +#!/bin/bash + +. $(dirname $0)/../include.rc +. $(dirname $0)/../volume.rc + +create_files () { + for i in {1..1000}; do + orig=$(printf %s/abc%04d $1 $i) + real=$(printf %s/src%04d $1 $i) + # Make sure lots of these have linkfiles. + echo "This is file $i" > $orig + mv $orig $real + done + sync +} + +move_files_inner () { + sfile=$M0/status_$(basename $1) + echo "running" > $sfile + for i in {1..1000}; do + src=$(printf %s/src%04d $1 $i) + dst=$(printf %s/dst%04d $1 $i) + mv $src $dst 2> /dev/null + done + echo "done" > $sfile +} + +move_files () { + move_files_inner $* & +} + +check_files () { + errors=0 + warnings=0 + for i in {1..1000}; do + if [ ! -f $(printf %s/dst%04d $1 $i) ]; then + if [ -f $(printf %s/src%04d $1 $i) ]; then + # We do hit this sometimes, though very rarely. + # It's a bug. It's just not *this* bug. + # Therefore, instead of allowing it to cause + # spurious test failures, we let it slide for + # now. Some day, when that other bug is fixed, + # I hope I remember to come back and strengthen + # this test accordingly. + echo "file $i didnt get moved" > /dev/stderr + #warnings=$((warnings+1)) + else + echo "file $i is MISSING" > /dev/stderr + errors=$((errors+1)) + fi + fi + done + if [ $((errors+warnings)) != 0 ]; then + : ls -l $1 > /dev/stderr + fi + return $errors +} + +cleanup; + +TEST glusterd +TEST pidof glusterd +TEST $CLI volume info; + +TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{1,2,3,4,5,6}; + +EXPECT "$V0" volinfo_field $V0 'Volume Name'; +EXPECT 'Created' volinfo_field $V0 'Status'; +EXPECT '6' brick_count $V0 + +TEST $CLI volume start $V0; +EXPECT 'Started' volinfo_field $V0 'Status'; + +## Mount FUSE with caching disabled (read-write) +TEST glusterfs --entry-timeout=0 --attribute-timeout=0 -s $H0 --volfile-id $V0 $M0; + +TEST create_files $M0 + +## Mount FUSE with caching disabled (read-write) again +TEST glusterfs --entry-timeout=0 --attribute-timeout=0 -s $H0 --volfile-id $V0 $M1; + +TEST move_files $M0 +TEST move_files $M1 + +# It's regrettable that renaming 1000 files might take more than 30 seconds, +# but on our test systems sometimes it does, so double the time from what we'd +# use otherwise. +EXPECT_WITHIN 300 "done" cat $M0/status_0 +EXPECT_WITHIN 300 "done" cat $M1/status_1 + +TEST umount $M0 +TEST umount $M1 +TEST glusterfs --entry-timeout=0 --attribute-timeout=0 -s $H0 --volfile-id $V0 $M0; +TEST check_files $M0 + +TEST $CLI volume stop $V0; +EXPECT 'Stopped' volinfo_field $V0 'Status'; + +TEST $CLI volume delete $V0; +TEST ! $CLI volume info $V0; + +cleanup; diff --git a/xlators/cluster/dht/src/dht-common.h b/xlators/cluster/dht/src/dht-common.h index 49fbd3878a0..83725f09712 100644 --- a/xlators/cluster/dht/src/dht-common.h +++ b/xlators/cluster/dht/src/dht-common.h @@ -183,6 +183,7 @@ struct dht_local { struct dht_rebalance_ rebalance; xlator_t *first_up_subvol; + gf_boolean_t added_link; }; typedef struct dht_local dht_local_t; diff --git a/xlators/cluster/dht/src/dht-rename.c b/xlators/cluster/dht/src/dht-rename.c index 35fedeaa7a4..7685ad88427 100644 --- a/xlators/cluster/dht/src/dht-rename.c +++ b/xlators/cluster/dht/src/dht-rename.c @@ -378,8 +378,9 @@ dht_rename_cleanup (call_frame_t *frame) if (dst_hashed != src_hashed && dst_hashed != src_cached) call_cnt++; - if (src_cached != dst_hashed) + if (local->added_link && (src_cached != dst_hashed)) { call_cnt++; + } local->call_cnt = call_cnt; @@ -395,10 +396,11 @@ dht_rename_cleanup (call_frame_t *frame) &local->loc, 0, NULL); } - if (src_cached != dst_hashed) { + if (local->added_link && (src_cached != dst_hashed)) { gf_log (this->name, GF_LOG_TRACE, "unlinking link %s => %s (%s)", local->loc.path, local->loc2.path, src_cached->name); + STACK_WIND (frame, dht_rename_unlink_cbk, src_cached, src_cached->fops->unlink, &local->loc2, 0, NULL); @@ -652,9 +654,15 @@ dht_rename_links_cbk (call_frame_t *frame, void *cookie, xlator_t *this, "link/file on %s failed (%s)", prev->this->name, strerror (op_errno)); local->op_ret = -1; - if (op_errno != ENOENT) + if (op_errno != ENOENT) { local->op_errno = op_errno; - } else { + + if (prev->this == local->src_cached) { + local->added_link = _gf_false; + } + } + } else if (local->src_cached == prev->this) { + /* merge of attr returned only from linkfile creation */ dht_iatt_merge (this, &local->stbuf, stbuf, prev->this); } @@ -767,6 +775,8 @@ dht_rename_create_links (call_frame_t *frame) gf_log (this->name, GF_LOG_TRACE, "link %s => %s (%s)", local->loc.path, local->loc2.path, src_cached->name); + + local->added_link = _gf_true; STACK_WIND (frame, dht_rename_links_cbk, src_cached, src_cached->fops->link, &local->loc, &local->loc2, NULL); -- cgit