Import wiredtiger: 15b4ef65bcd4dac075a5aa3fd03f013abc11c21e from branch mongodb-4.6

ref: dbcb682272..15b4ef65bc
for: 4.5.1

WT-6532       Consider update structure overhead in split length calculation
WT-6615       Initialize last_upd where it is actually used
WT-6623       Set the connection level file id in recovery file scan
WT-6635       Disable mix and column filetype test

Author: lukech

src/third_party/wiredtiger/import.data

@@ -2,5 +2,5 @@
     "vendor": "wiredtiger",
     "github": "wiredtiger/wiredtiger.git",
     "branch": "mongodb-4.6",
-    "commit": "dbcb6822729ee16eceeee28a8be7ce7aac6d8835"
+    "commit": "15b4ef65bcd4dac075a5aa3fd03f013abc11c21e"
 }

src/third_party/wiredtiger/src/include/reconcile.i

@@ -217,10 +217,12 @@ __wt_rec_need_split(WT_RECONCILE *r, size_t len)
      * updates) associated with the page. If there are barely any items on the page, then it's not
      * worth splitting. We also want to temper this effect to avoid in-memory updates from
      * dominating the calculation and causing excessive splitting. Therefore, we'll limit the impact
-     * to a tenth of the cache usage occupied by those updates.
+     * to a tenth of the cache usage occupied by those updates. For small updates the overhead of
+     * the update structure could skew the calculation, so we subtract the overhead before
+     * considering the cache usage by the updates.
      */
     if (r->page->type == WT_PAGE_ROW_LEAF && page_items > WT_REC_SPLIT_MIN_ITEMS_USE_MEM)
-        len += r->supd_memsize / 10;
+        len += (r->supd_memsize - (r->supd_next * WT_UPDATE_SIZE)) / 10;
 
     /* Check for the disk image crossing a boundary. */
     return (WT_CHECK_CROSSING_BND(r, len));

src/third_party/wiredtiger/src/reconcile/rec_visibility.c

@@ -440,8 +440,6 @@ __wt_rec_upd_select(WT_SESSION_IMPL *session, WT_RECONCILE *r, WT_INSERT *ins, v
                 }
 
                 WT_ASSERT(session, upd->next == NULL || upd->next->txnid != WT_TXN_ABORTED);
-                if (upd->next == NULL)
-                    last_upd = upd;
                 upd_select->upd = upd = upd->next;
 
                 /*
@@ -475,9 +473,26 @@ __wt_rec_upd_select(WT_SESSION_IMPL *session, WT_RECONCILE *r, WT_INSERT *ins, v
             /* The beginning of the validity window is the selected update's time point. */
             WT_TIME_WINDOW_SET_START(select_tw, upd);
         else if (select_tw->stop_ts != WT_TS_NONE || select_tw->stop_txn != WT_TXN_NONE) {
-            /* If we only have a tombstone in the update list, we must have an ondisk value. */
-            WT_ASSERT(session,
-              vpack != NULL && tombstone != NULL && !vpack->tw.prepare && last_upd->next == NULL);
+            /*
+             * We only have a tombstone on the update list or all the updates are from the same
+             * transaction.
+             */
+            WT_ASSERT(session, tombstone != NULL);
+
+            /* We must have an ondisk value and it can't be a prepared update. */
+            WT_ASSERT(session, vpack != NULL && vpack->type != WT_CELL_DEL && !vpack->tw.prepare);
+
+            /* Move the pointer to the last update on the update chain. */
+            for (last_upd = tombstone; last_upd->next != NULL; last_upd = last_upd->next)
+                /*
+                 * Tombstone is the only non-aborted update on the update chain or all the updates
+                 * are from the same transaction.
+                 */
+                WT_ASSERT(session,
+                  last_upd->next->txnid == WT_TXN_ABORTED ||
+                    (last_upd->next->txnid == tombstone->txnid &&
+                      last_upd->next->start_ts == tombstone->start_ts));
+
             /*
              * It's possible to have a tombstone as the only update in the update list. If we
              * reconciled before with only a single update and then read the page back into cache,
@@ -507,6 +522,10 @@ __wt_rec_upd_select(WT_SESSION_IMPL *session, WT_RECONCILE *r, WT_INSERT *ins, v
                     WT_TIME_WINDOW_SET_START(select_tw, last_upd->next);
                 } else {
                     /*
+                     * It's possible that onpage value is not appended if the tombstone becomes
+                     * globally visible because the oldest transaction id or the oldest timestamp is
+                     * moved concurrently.
+                     *
                      * If the tombstone is aborted concurrently, we should still have appended the
                      * onpage value.
                      */

src/third_party/wiredtiger/src/txn/txn_recover.c

@@ -558,6 +558,13 @@ __recovery_file_scan(WT_RECOVERY *r)
         WT_RET(__recovery_setup_file(r, uri, config));
     }
     WT_RET_NOTFOUND_OK(ret);
+
+    /*
+     * Set the connection level file id tracker, as such upon creation of a new file we'll begin
+     * from the latest file id.
+     */
+    S2C(r->session)->next_file_id = r->max_fileid;
+
     return (0);
 }
 
@@ -683,7 +690,6 @@ __wt_txn_recover(WT_SESSION_IMPL *session, const char *cfg[])
          * the files.
          */
         metafile = &r.files[WT_METAFILE_ID];
-        conn->next_file_id = r.max_fileid;
 
         if (FLD_ISSET(conn->log_flags, WT_CONN_LOG_ENABLED) && WT_IS_MAX_LSN(&metafile->ckpt_lsn) &&
           !WT_IS_MAX_LSN(&r.max_ckpt_lsn))
@@ -814,8 +820,6 @@ __wt_txn_recover(WT_SESSION_IMPL *session, const char *cfg[])
         ret = 0;
     WT_ERR(ret);
 
-    conn->next_file_id = r.max_fileid;
-
 done:
     WT_ERR(__recovery_set_checkpoint_timestamp(&r));
     WT_ERR(__recovery_set_oldest_timestamp(&r));

src/third_party/wiredtiger/test/evergreen.yml

@@ -1914,15 +1914,17 @@ tasks:
         vars:
           # Don't use diagnostic - this test looks for timing problems that are more likely to occur without it
           posix_configure_flags: --enable-strict
-      - func: "checkpoint test"
-        vars:
-          checkpoint_args: -t m -n 1000000 -k 5000000 -C cache_size=100MB
+      # Temporarily disable mix and column file type
+      # FIXME after WT-6608
+      #- func: "checkpoint test"
+      #  vars:
+      #    checkpoint_args: -t m -n 1000000 -k 5000000 -C cache_size=100MB
+      #- func: "checkpoint test"
+      #  vars:
+      #    checkpoint_args: -t c -n 1000000 -k 5000000 -C cache_size=100MB
       - func: "checkpoint test"
         vars:
           checkpoint_args: -t r -n 1000000 -k 5000000 -C cache_size=100MB
-      - func: "checkpoint test"
-        vars:
-          checkpoint_args: -t c -n 1000000 -k 5000000 -C cache_size=100MB
 
   - name: coverage-report
     commands: