Conversation

Notices

1.  (mint@eientei.org)'s status on Thursday, 16-Feb-2023 08:23:45 JST 
   Forgive me mama, I've used make install.
   •  (mint@eientei.org)'s status on Thursday, 16-Feb-2023 08:23:37 JST 

     in reply to
     Even with binary format, you need --disable-triggers for it to not sperg out with each copied column.
   •  (mint@eientei.org)'s status on Thursday, 16-Feb-2023 08:23:39 JST 

     in reply to
     pg_dumpall does not support custom format. What a bummer.
   •  (mint@eientei.org)'s status on Thursday, 16-Feb-2023 08:23:40 JST 

     in reply to
     psql:data.sql:2181309: ERROR: function activity_visibility(character varying, character varying[], jsonb) does not exist LINE 1: visibility_new := activity_visibility(NEW.actor, NEW.recipie... ^ HINT: No function matches the given name and argument types. You might need to add explicit type casts. QUERY: visibility_new := activity_visibility(NEW.actor, NEW.recipients, NEW.data) CONTEXT: PL/pgSQL function public.update_status_visibility_counter_cache() line 15 at assignment COPY activities, line 1: "00000185-47a3-fb00-d73b-6f576a0e0000 {"cc": ["https://artisan.chat/users/Curator", "https://chaos.so..." COPY 0 COPY 81922 COPY 0 COPY 589 COPY 0 psql:data.sql:2263869: ERROR: insert or update on table "bookmarks" violates foreign key constraint "bookmarks_activity_id_fkey" DETAIL: Key (activity_id)=(00000185-e7ca-4951-847f-1d724ce70000) is not present in table "activities".

     Not a good sign.

   •  (mint@eientei.org)'s status on Thursday, 16-Feb-2023 08:23:41 JST 

     in reply to
     Looks like the activity index bottleneck is still there. Wasted an hour, so let's interrupt this, recreate the cluster with pgtune settings and dump schema and data separately.
   •  (mint@eientei.org)'s status on Thursday, 16-Feb-2023 08:23:42 JST 

     in reply to
     Forgor 💀 to use pg_tune to speed up the database before going with the dump restore. Hopefully it won't cause any issues.
   •  (mint@eientei.org)'s status on Thursday, 16-Feb-2023 08:23:43 JST 

     in reply to
     And I'm not sure if there's enough space on root to keep both the old and new cluster. Will move all the old containers to the terabyte slab.
   •  (mint@eientei.org)'s status on Thursday, 16-Feb-2023 08:24:12 JST 

     in reply to
     Takes quite a while, but at least I can see it's doing COPY and not CREATE INDEX. Feeling rather sleepy right now, might have to finish the job tomorrow.