pg_dump --exclude-schema=foo
We have discovered an interesting locking scenario with Slony-I that
is pointing to a use for the ability to exclude certain schemas from
pg_dump.
The situation is that when a "full" pg_dump kicks off, a Slony-I
"create sync" event, which expects to "LOCK slony_schema.sl_event;",
is blocked from getting the exclusive lock that it desires.
Everything else then falls behind that, as they are waiting for the
lock to get dropped off.
The solution is easy enough, in that (at present) the interesting data
is in a single database schema.
Running...
pg_dump --schema=ourdata
does the trick.
What would be nice would be the ability to consciously _exclude_
schemas, where the nice way would be --exclude-schema=_slony_schema
If someone has some "round tuits," I'll have to catch up on sleep...
--
(format nil "~S@~S" "cbbrowne" "ntlug.org")
http://www3.sympatico.ca/cbbrowne/languages.html
Save your burned out bulbs for me, I'm building my own dark room.
Chris Browne <cbbrowne@acm.org> writes:
We have discovered an interesting locking scenario with Slony-I that
is pointing to a use for the ability to exclude certain schemas from
pg_dump.
The situation is that when a "full" pg_dump kicks off, a Slony-I
"create sync" event, which expects to "LOCK slony_schema.sl_event;",
is blocked from getting the exclusive lock that it desires.
Perhaps this should be seen as a Slony bug. Does it really need an
AccessExclusive lock, or would an Exclusive lock do?
regards, tom lane
Quoth tgl@sss.pgh.pa.us (Tom Lane):
Chris Browne <cbbrowne@acm.org> writes:
We have discovered an interesting locking scenario with Slony-I that
is pointing to a use for the ability to exclude certain schemas from
pg_dump.The situation is that when a "full" pg_dump kicks off, a Slony-I
"create sync" event, which expects to "LOCK slony_schema.sl_event;",
is blocked from getting the exclusive lock that it desires.Perhaps this should be seen as a Slony bug. Does it really need an
AccessExclusive lock, or would an Exclusive lock do?
You could be right; that's still To Be Determined.
It nonetheless appears like a useful idea in general to be able to
specify inclusion/exclusion of multiple schemas.
--
(reverse (concatenate 'string "gro.mca" "@" "enworbbc"))
http://www.ntlug.org/~cbbrowne/languages.html
"You can only examine 10 levels of pushdown, because that's all the
fingers you have to stick in the listing."
-- Anonymous programmer - "TOPS-10 Crash Analysis Guide"