Mark Stosberg <mark at summersault.com> writes:

> I just ran "slonik_uninstall_nodes" and got an error about a missing
> table. After I thought I fixed it, I got this response when I ran it again:
>
> root at sap# slonik_uninstall_nodes | slonik
> <stdin>:4: PGRES_FATAL_ERROR select "_ff".uninstallNode();  - ERROR:
> schema "_foo" does not exist
>
> What happened was the table error was on node 1, which is uninstalled
> /after/ node2. I think node 2 was successfully uninstalled on the first
> try.
>
> So when I re-ran the script, I think it's bombing out because node2 no
> longer exists.
>
> It would be more clearer and helpful if the feedback looked like this
> instead:
>
>  Node 2 appears to already be uninstalled. Skipping.
>  Node 1:<stdin>:4: ... (node 1 errors, if any, here).

OK, here's a patch for this...

cbbrowne at dba2:Slony-I/slony1-HEAD-pristine/src/slonik> cvs diff -u slonik.c
Index: slonik.c
===================================================================
RCS file: /usr/local/cvsroot/slony1/slony1-engine/src/slonik/slonik.c,v
retrieving revision 1.71
diff -c -u -r1.71 slonik.c
cvs diff: conflicting specifications of output style
--- slonik.c	12 Dec 2006 14:54:48 -0000	1.71
+++ slonik.c	30 Jan 2007 17:21:33 -0000
@@ -3034,6 +3034,7 @@
 				 stmt->hdr.script->clustername);
 	if (db_exec_command((SlonikStmt *) stmt, adminfo1, &query) < 0)
 	{
+    printf("Failed to run uninstallNode() for node %d\n", stmt->no_id);
 		dstring_free(&query);
 		return -1;
 	}

Output of this:
cbbrowne at dba2:Slony-I/slony1-HEAD-pristine/src/slonik> ./slonik slonik.script
slonik.script:4: PGRES_FATAL_ERROR select "_slony_regress1".uninstallNode();  - ERROR:  schema "_slony_regress1" does not exist
Failed to run uninstallNode() for node 2

This seems a reasonable change to commit, and isn't version specific.
I'll apply to 1.1, 1.2, and HEAD.
-- 
select 'cbbrowne' || '@' || 'ca.afilias.info';
<http://dba2.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)