pg.ddx.io pgsql-admin@postgresql.org mailing list archive
help / color / mirror / Atom feedUpgrade failure attempting to go to 12.0
2+ messages / 2 participants
[nested] [flat]
* Upgrade failure attempting to go to 12.0
@ 2019-10-23 02:48 Karl Denninger <karl@denninger.net>
0 siblings, 1 reply; 2+ messages in thread
From: Karl Denninger @ 2019-10-23 02:48 UTC (permalink / raw)
To: pgsql-admin@lists.postgresql.org
Coming from 10.1.
I think I've seen this before, but I can't figure out where the issue
is. The original database dates to way, way back when contrib/tsearch2
was a loaded module instead of the functionality being part of the base
system, and I suspect vestigal pieces are involved.
When I try to run the upgrade I get this failure when pg_upgrade gets to
template1:
pg_restore: creating FUNCTION "public.pgp_sym_decrypt("bytea", "text")"
pg_restore: creating FUNCTION "public.pgp_sym_decrypt("bytea", "text",
"text")"
pg_restore: creating FUNCTION "public.pgp_sym_decrypt_bytea("bytea",
"text")"
pg_restore: creating FUNCTION "public.pgp_sym_decrypt_bytea("bytea",
"text", "te
xt")"
pg_restore: creating FUNCTION "public.pgp_sym_encrypt("text", "text")"
pg_restore: creating FUNCTION "public.pgp_sym_encrypt("text", "text",
"text")"
pg_restore: creating FUNCTION "public.pgp_sym_encrypt_bytea("bytea",
"text")"
pg_restore: creating FUNCTION "public.pgp_sym_encrypt_bytea("bytea",
"text", "te
xt")"
pg_restore: creating FUNCTION "public.querytree("tsquery")"
pg_restore: creating FUNCTION "public.setweight("tsvector", "char")"
pg_restore: creating FUNCTION "public.strip("tsvector")"
pg_restore: creating FUNCTION "public.ts_debug("text")"
pg_restore: creating FUNCTION "public.tsq_mcontained("tsquery", "tsquery")"
pg_restore: creating FUNCTION "public.tsq_mcontains("tsquery", "tsquery")"
pg_restore: creating FUNCTION "public.tsquery_and("tsquery", "tsquery")"
pg_restore: creating FUNCTION "public.tsquery_not("tsquery")"
pg_restore: creating FUNCTION "public.tsquery_or("tsquery", "tsquery")"
pg_restore: creating OPERATOR FAMILY "public.gin_tsvector_ops"
pg_restore: creating OPERATOR CLASS "public.gin_tsvector_ops"
pg_restore: creating OPERATOR FAMILY "public.gist_tp_tsquery_ops"
pg_restore: creating OPERATOR CLASS "public.gist_tp_tsquery_ops"
pg_restore: while PROCESSING TOC:
pg_restore: from TOC entry 1579; 2616 17787 OPERATOR CLASS
gist_tp_tsquery_ops p
gsql
pg_restore: error: could not execute query: ERROR: function
gtsquery_decompress
(internal) does not exist
Command was: CREATE OPERATOR CLASS "public"."gist_tp_tsquery_ops"
FOR TYPE "tsquery" USING "gist" FAMILY "public"."gist_tp_tsquery_ops" AS
STORAGE bigint ,
OPERATOR 7 @>("tsquery","tsquery") ,
OPERATOR 8 <@("tsquery","tsquery") ,
FUNCTION 1 ("tsquery", "tsquery")
"gtsquery_consistent"("internal","internal
",integer,"oid","internal") ,
FUNCTION 2 ("tsquery", "tsquery")
"gtsquery_union"("internal","internal") ,
FUNCTION 3 ("tsquery", "tsquery") "gtsquery_compress"("internal") ,
FUNCTION 4 ("tsquery", "tsquery") "gtsquery_decompress"("internal") ,
FUNCTION 5 ("tsquery", "tsquery")
"gtsquery_penalty"("internal","internal","
internal") ,
FUNCTION 6 ("tsquery", "tsquery")
"gtsquery_picksplit"("internal","internal"
) ,
FUNCTION 7 ("tsquery", "tsquery")
"gtsquery_same"(bigint,bigint,"internal");
I *think* this is from vestigal parts of the old tsearch2
components..... but I can't find them in the old template1 database (am
I looking in the right place?)
I suspect these are also in all the other databases in the cluster, and
thus I need to kill them for each of those... but I have to find 'em first!
Thanks in advance (please be specific -- this one has me stumped!)
--
Karl Denninger
karl@denninger.net <mailto:karl@denninger.net>
/The Market Ticker/
/[S/MIME encrypted email preferred]/
Attachments:
smime.p7s (application/pkcs7-signature, 4.8K) download
^ permalink raw reply [nested|flat] 2+ messages in thread
* Re: Upgrade failure attempting to go to 12.0
@ 2019-10-23 04:38 Tom Lane <tgl@sss.pgh.pa.us>
parent: Karl Denninger <karl@denninger.net>
0 siblings, 0 replies; 2+ messages in thread
From: Tom Lane @ 2019-10-23 04:38 UTC (permalink / raw)
To: Karl Denninger <karl@denninger.net>; +Cc: pgsql-admin@lists.postgresql.org
Karl Denninger <karl@denninger.net> writes:
> I think I've seen this before, but I can't figure out where the issue
> is. The original database dates to way, way back when contrib/tsearch2
> was a loaded module instead of the functionality being part of the base
> system, and I suspect vestigal pieces are involved.
Yeah, that's what it looks like. This:
> Command was: CREATE OPERATOR CLASS "public"."gist_tp_tsquery_ops"
is part of the old tsearch2 stuff, and you'll need to remove it from
the v10 installation before you can upgrade.
What I can't tell from this evidence is whether you'd ever converted
tsearch2 into an actual extension --- if you had, then "drop extension"
would be enough to get rid of it. Otherwise, you need to consult a copy
of the old "uninstall_tsearch2.sql" script to find out what you ought to
get rid of.
If you have user tables or indexes that depend on the old tsearch2
objects, things get more complex :-( --- you'd want to update those
to depend on the newer core datatype and index opclasses before
you clean out the old ones.
regards, tom lane
^ permalink raw reply [nested|flat] 2+ messages in thread
end of thread, other threads:[~2019-10-23 04:38 UTC | newest]
Thread overview: 2+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2019-10-23 02:48 Upgrade failure attempting to go to 12.0 Karl Denninger <karl@denninger.net>
2019-10-23 04:38 ` Tom Lane <tgl@sss.pgh.pa.us>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox