I can't see any good reason for DropRelFileNodeBuffers to be issuing
authorTom Lane
Mon, 6 Sep 2004 17:31:32 +0000 (17:31 +0000)
committerTom Lane
Mon, 6 Sep 2004 17:31:32 +0000 (17:31 +0000)
FATAL when it detects a nonzero reference count.  Reduce to ERROR.

src/backend/storage/buffer/bufmgr.c

index b9ba90cd8ecdc970630c1ca9e764ce7646b14bfd..b98193ad38c74ddd5adf7de0f89a4a35d38e3215 100644 (file)
@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *   $PostgreSQL: pgsql/src/backend/storage/buffer/bufmgr.c,v 1.176 2004/08/31 16:13:06 tgl Exp $
+ *   $PostgreSQL: pgsql/src/backend/storage/buffer/bufmgr.c,v 1.177 2004/09/06 17:31:32 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -1129,7 +1129,7 @@ DropRelFileNodeBuffers(RelFileNode rnode, bool istemp,
                bufHdr->tag.blockNum >= firstDelBlock)
            {
                if (LocalRefCount[i] != 0)
-                   elog(FATAL, "block %u of %u/%u/%u is still referenced (local %u)",
+                   elog(ERROR, "block %u of %u/%u/%u is still referenced (local %u)",
                         bufHdr->tag.blockNum,
                         bufHdr->tag.rnode.spcNode,
                         bufHdr->tag.rnode.dbNode,
@@ -1172,7 +1172,7 @@ recheck:
             * There should be no pin on the buffer.
             */
            if (bufHdr->refcount != 0)
-               elog(FATAL, "block %u of %u/%u/%u is still referenced (private %d, global %u)",
+               elog(ERROR, "block %u of %u/%u/%u is still referenced (private %d, global %u)",
                     bufHdr->tag.blockNum,
                     bufHdr->tag.rnode.spcNode,
                     bufHdr->tag.rnode.dbNode,