Test bug similar to 307c1731c1

.gitea/workflows/test.yml

@@ -550,6 +550,24 @@ jobs:
           echo ""
         done
 
+  test_heal_imm_ec:
+    runs-on: ubuntu-latest
+    needs: build
+    container: ${{env.TEST_IMAGE}}:${{github.sha}}
+    steps:
+    - name: Run test
+      id: test
+      timeout-minutes: 10
+      run: IMMEDIATE_COMMIT=1 SCHEME=ec /root/vitastor/tests/test_heal.sh
+    - name: Print logs
+      if: always() && steps.test.outcome == 'failure'
+      run: |
+        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
+          echo "-------- $i --------"
+          cat $i
+          echo ""
+        done
+
   test_scrub:
     runs-on: ubuntu-latest
     needs: build

src/blockstore_flush.cpp

@@ -536,7 +536,19 @@ resume_1:
                 return false;
             }
             // zero out old metadata entry
+            {
+                clean_disk_entry *old_entry = (clean_disk_entry*)((uint8_t*)meta_old.buf + meta_old.pos*bs->dsk.clean_entry_size);
+                if (old_entry->oid.inode != 0 && old_entry->oid != cur.oid)
+                {
+                    printf("Fatal error (metadata corruption or bug): tried to wipe metadata entry %lu (%lx:%lx v%lu) as old location of %lx:%lx\n",
+                        old_clean_loc >> bs->dsk.block_order, old_entry->oid.inode, old_entry->oid.stripe,
+                        old_entry->version, cur.oid.inode, cur.oid.stripe);
+                    exit(1);
+                }
+            }
             memset((uint8_t*)meta_old.buf + meta_old.pos*bs->dsk.clean_entry_size, 0, bs->dsk.clean_entry_size);
+            if (meta_old.sector != meta_new.sector)
+            {
                 await_sqe(15);
                 data->iov = (struct iovec){ meta_old.buf, bs->dsk.meta_block_size };
                 data->callback = simple_callback_w;
@@ -545,6 +557,7 @@ resume_1:
                 );
                 wait_count++;
             }
+        }
         if (has_delete)
         {
             clean_disk_entry *new_entry = (clean_disk_entry*)((uint8_t*)meta_new.buf + meta_new.pos*bs->dsk.clean_entry_size);

src/blockstore_init.cpp

@@ -790,7 +790,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                     unstab = unstab < ov.version ? ov.version : unstab;
                     if (je->type == JE_SMALL_WRITE_INSTANT)
                     {
-                        bs->mark_stable(ov, true);
+                        bs->mark_stable(ov);
                     }
                 }
             }
@@ -890,7 +890,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                     unstab = unstab < ov.version ? ov.version : unstab;
                     if (je->type == JE_BIG_WRITE_INSTANT)
                     {
-                        bs->mark_stable(ov, true);
+                        bs->mark_stable(ov);
                     }
                 }
             }
@@ -904,7 +904,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                     .oid = je->stable.oid,
                     .version = je->stable.version,
                 };
-                bs->mark_stable(ov, true);
+                bs->mark_stable(ov);
             }
             else if (je->type == JE_ROLLBACK)
             {
@@ -961,7 +961,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                     bs->journal.used_sectors[proc_pos]++;
                     // Deletions are treated as immediately stable, because
                     // "2-phase commit" (write->stabilize) isn't sufficient for them anyway
-                    bs->mark_stable(ov, true);
+                    bs->mark_stable(ov);
                 }
                 // Ignore delete if neither preceding dirty entries nor the clean one are present
             }

tests/run_tests.sh

@@ -46,6 +46,7 @@ SCHEME=xor ./test_write.sh
 
 PG_SIZE=2 ./test_heal.sh
 SCHEME=ec ./test_heal.sh
+IMMEDIATE_COMMIT=1 SCHEME=ec ./test_heal.sh
 
 ./test_scrub.sh
 ZERO_OSD=2 ./test_scrub.sh

tests/test_heal.sh

@@ -12,7 +12,7 @@ PG_COUNT=32
 . `dirname $0`/run_3osds.sh
 check_qemu
 
-IMG_SIZE=960
+IMG_SIZE=128
 
 $ETCDCTL put /vitastor/config/inode/1/1 '{"name":"testimg","size":'$((IMG_SIZE*1024*1024))'}'
 
@@ -22,32 +22,28 @@ LD_PRELOAD="build/src/libfio_vitastor.so" \
 
 kill_osds()
 {
+    for i in {1..60}; do
         sleep 5
-
-    kill -9 $OSD1_PID
-    $ETCDCTL del /vitastor/osd/state/1
-
-    for i in 2 3 4 5 6 7; do
-        sleep 15
-        echo Killing OSD $i and starting OSD $((i-1))
-        p=OSD${i}_PID
-        kill -9 ${!p}
-        $ETCDCTL del /vitastor/osd/state/$i
-        start_osd $((i-1))
-        sleep 15
+        K=$((1 + RANDOM % OSD_COUNT))
+        eval "kill -9 \$OSD${K}_PID"
+        $ETCDCTL del /vitastor/osd/state/$K
+        start_osd $K
     done
-
-    sleep 5
-    start_osd 7
-
-    sleep 5
 }
 
 kill_osds &
 
 LD_PRELOAD="build/src/libfio_vitastor.so" \
-    fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=4k -direct=1 -iodepth=16 -fsync=256 -rw=randwrite \
-        -refill_buffers=1 -mirror_file=./testdata/mirror.bin -etcd=$ETCD_URL -image=testimg -loops=10 -runtime=120
+    fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=512k -direct=1 -iodepth=8 -fsync=256 -rw=write \
+        -refill_buffers=1 -mirror_file=./testdata/mirror.bin -etcd=$ETCD_URL -image=testimg -loops=1000 -runtime=120 &>/dev/null &
+
+LD_PRELOAD="build/src/libfio_vitastor.so" \
+    fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=512k -direct=1 -iodepth=8 -fsync=256 -rw=write \
+        -refill_buffers=1 -mirror_file=./testdata/mirror.bin -etcd=$ETCD_URL -image=testimg -loops=1000 -runtime=120 &>/dev/null &
+
+LD_PRELOAD="build/src/libfio_vitastor.so" \
+    fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=512k -direct=1 -iodepth=8 -fsync=256 -rw=write \
+        -refill_buffers=1 -mirror_file=./testdata/mirror.bin -etcd=$ETCD_URL -image=testimg -loops=1000 -runtime=120
 
 qemu-img convert -S 4096 -p \
     -f raw "vitastor:etcd_host=127.0.0.1\:$ETCD_PORT/v3:image=testimg" \