Re: crash-recovery replay of CREATE TABLESPACE is broken in HEAD

Bruce Momjian wrote:
> Bruce Momjian wrote:
> > > The attached patch does as suggested.  I added the recovery code to the
> > > create tablespace function so I didn't have to duplicate all the code
> > > that computes the path names.
> > >
> > > Attached.
> >
> > Uh, another question.  Looking at the createdb recovery, I see:
> >
> >         /*
> >          * Our theory for replaying a CREATE is to forcibly drop the target
> >          * subdirectory if present, then re-copy the source data. This may be
> >          * more work than needed, but it is simple to implement.
> >          */
> >         if (stat(dst_path, &st) == 0 && S_ISDIR(st.st_mode))
> >         {
> >             if (!rmtree(dst_path, true))
> >                 ereport(WARNING,
> >                         (errmsg("some useless files may be left behind in old database directory \"%s\"",
> >                                 dst_path)));
> >         }
> >
> > Should I be using rmtree() on the mkdir target?
> >
> > Also, the original tablespace recovery code did not drop the symlink
> > first.  I assume that was not a bug only because we don't support moving
> > tablespaces:
>
> For consistency with CREATE DATABASE recovery and for reliablity, I
> coded the rmtree() call instead.  Patch attached.

Attached patch applied to HEAD and 9.0.   9.0 open item moved to
completed.

--
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

  + None of us is going to be here forever. +
Index: src/backend/commands/dbcommands.c
===================================================================
RCS file: /cvsroot/pgsql/src/backend/commands/dbcommands.c,v
retrieving revision 1.235
diff -c -c -r1.235 dbcommands.c
*** src/backend/commands/dbcommands.c    26 Feb 2010 02:00:38 -0000    1.235
--- src/backend/commands/dbcommands.c    20 Jul 2010 18:11:06 -0000
***************
*** 1908,1913 ****
--- 1908,1914 ----
          if (stat(dst_path, &st) == 0 && S_ISDIR(st.st_mode))
          {
              if (!rmtree(dst_path, true))
+                 /* If this failed, copydir() below is going to error. */
                  ereport(WARNING,
                          (errmsg("some useless files may be left behind in old database directory \"%s\"",
                                  dst_path)));
Index: src/backend/commands/tablespace.c
===================================================================
RCS file: /cvsroot/pgsql/src/backend/commands/tablespace.c,v
retrieving revision 1.77
diff -c -c -r1.77 tablespace.c
*** src/backend/commands/tablespace.c    18 Jul 2010 04:47:46 -0000    1.77
--- src/backend/commands/tablespace.c    20 Jul 2010 18:11:07 -0000
***************
*** 562,567 ****
--- 562,586 ----
                           location)));
      }

+     if (InRecovery)
+     {
+         struct stat st;
+
+         /*
+          * Our theory for replaying a CREATE is to forcibly drop the target
+          * subdirectory if present, and then recreate it. This may be
+          * more work than needed, but it is simple to implement.
+          */
+         if (stat(location_with_version_dir, &st) == 0 && S_ISDIR(st.st_mode))
+         {
+             if (!rmtree(location_with_version_dir, true))
+                 /* If this failed, mkdir() below is going to error. */
+                 ereport(WARNING,
+                         (errmsg("some useless files may be left behind in old database directory \"%s\"",
+                                 location_with_version_dir)));
+         }
+     }
+
      /*
       * The creation of the version directory prevents more than one tablespace
       * in a single location.
***************
*** 580,585 ****
--- 599,614 ----
                              location_with_version_dir)));
      }

+     /* Remove old symlink in recovery, in case it points to the wrong place */
+     if (InRecovery)
+     {
+         if (unlink(linkloc) < 0 && errno != ENOENT)
+             ereport(ERROR,
+                     (errcode_for_file_access(),
+                      errmsg("could not remove symbolic link \"%s\": %m",
+                             linkloc)));
+     }
+
      /*
       * Create the symlink under PGDATA
       */