LU-11226 flr: mirror resync regression

There is a glitch in the lfs mirror resync tool in commit
0e5c12ac29a9622e8ca05d5e39cd5e2a721ace93, resync write needs to
restricted to the component's extent.

Signed-off-by: Bobi Jam <bobijam@whamcloud.com>
Change-Id: Ifbd3f16b2f621407b31c7fe37ce9745de48fcc99
Reviewed-on: https://review.whamcloud.com/32968
Tested-by: Jenkins
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>

diff --git a/lustre/tests/sanity-flr.sh b/lustre/tests/sanity-flr.sh
index fd5f5c6..b732065 100644 (file)
--- a/lustre/tests/sanity-flr.sh
+++ b/lustre/tests/sanity-flr.sh
@@ -1477,9 +1477,11 @@ test_41() {
        # write data in [0, 2M)
        dd if=/dev/zero of=$tf bs=1M count=2 conv=notrunc ||
                error "writing $tf failed"
-       dd if=/dev/zero of=$tf-1 bs=1M count=4 conv=notrunc ||
+       dd if=/dev/urandom of=$tf-1 bs=1M count=4 conv=notrunc ||
                error "writing $tf-1 failed"
 
+       local sum0=$(cat $tf-1 | md5sum | cut -f 1 -d' ')
+
        echo " **verify files be WRITE_PENDING"
        verify_flr_state $tf "wp"
        verify_flr_state $tf-1 "wp"
@@ -1494,6 +1496,14 @@ test_41() {
        echo " **full resync"
        $LFS mirror resync $tf $tf-1 || error "mirror resync $tf $tf-1 failed"
 
+       echo " **verify $tf-1 data consistency in all mirrors"
+       local sum
+       for i in 1 2 3; do
+               sum=$(mirror_io dump -i $i $tf-1 | md5sum | cut -f 1 -d' ')
+               [ "$sum" = "$sum0" ] ||
+                       error "$i: mismatch: $sum vs. $sum0"
+       done
+
        echo " **verify files be RDONLY"
        verify_flr_state $tf "ro"
        verify_flr_state $tf-1 "ro"
@@ -1995,9 +2005,13 @@ resync_file_200() {
 
                echo -n "resync file $tf with '$cmd' .."
 
-               $lock_taken && flock -x 200
-               $cmd $tf &> /dev/null && echo "done" || echo "failed"
-               $lock_taken && flock -u 200
+               if [[ $lock_taken = "true" ]]; then
+                       flock -x 200 -c "$cmd $tf &> /dev/null" &&
+                               echo "done" || echo "failed"
+                       flock -u 200
+               else
+                       $cmd $tf &> /dev/null && echo "done" || echo "failed"
+               fi
 
                sleep 0.$((RANDOM % 8 + 1))
        done

diff --git a/lustre/utils/liblustreapi_layout.c b/lustre/utils/liblustreapi_layout.c
index a282d7a..a0905d7 100644 (file)
--- a/lustre/utils/liblustreapi_layout.c
+++ b/lustre/utils/liblustreapi_layout.c
@@ -2601,15 +2601,27 @@ int llapi_mirror_resync_many(int fd, struct llapi_layout *layout,
 
                for (i = 0; i < comp_size; i++) {
                        ssize_t written;
+                       off_t pos2 = pos;
+                       size_t to_write2 = to_write;
 
                        /* skip non-overlapped component */
-                       if (pos > comp_array[i].lrc_end ||
-                           pos + to_write < comp_array[i].lrc_start)
+                       if (pos >= comp_array[i].lrc_end ||
+                           pos + to_write <= comp_array[i].lrc_start)
                                continue;
 
+                       if (pos < comp_array[i].lrc_start)
+                               pos2 = comp_array[i].lrc_start;
+
+                       to_write2 -= pos2 - pos;
+
+                       if ((pos + to_write) > comp_array[i].lrc_end)
+                               to_write2 -= pos + to_write -
+                                            comp_array[i].lrc_end;
+
                        written = llapi_mirror_write(fd,
-                                       comp_array[i].lrc_mirror_id, buf,
-                                       to_write, pos);
+                                       comp_array[i].lrc_mirror_id,
+                                       buf + pos2 - pos,
+                                       to_write2, pos2);
                        if (written < 0) {
                                /**
                                 * this component is not written successfully,
@@ -2624,7 +2636,7 @@ int llapi_mirror_resync_many(int fd, struct llapi_layout *layout,
                                comp_array[i].lrc_synced = true;
                                continue;
                        }
-                       assert(written == to_write);
+                       assert(written == to_write2);
                }
 
                pos += bytes_read;