Re: pg_upgrade from Postgresql-12 to Postgresql-13 fails with "Creating dump of database schemas postgres *failure*"

Hi, Laurenz, 

I tried out both `pg_dumpall` and `pg_dump`.  

`pg_dumpall` went by alphbetical order when dumping the database. It always succeeds the first, but fails at the second database (`ankara` is the second) onwards with the following thread:

```
postgres@ual:~$ pg_dumpall -U postgres | gzip > /bnas/pgdumps/$(date +"%Y_%m_%d")_pg_bck.gz
pg_dump: WARNING:  terminating connection because of crash of another server process
DETAIL:  The postmaster has commanded this server process to roll back the current transaction and exit, because another server process exited abnormally and possibly corrupted shared memory.
HINT:  In a moment you should be able to reconnect to the database and repeat your command.
pg_dump: error: query failed: server closed the connection unexpectedly
        This probably means the server terminated abnormally
        before or while processing the request.
pg_dump: error: query was: SELECT at.attname, (SELECT pg_catalog.array_agg(acl ORDER BY row_n) FROM (SELECT acl, row_n FROM pg_catalog.unnest(coalesce(at.attacl,pg_catalog.acldefault('c',c.relowner))) WITH ORDINALITY AS perm(acl,row_n) WHERE NOT EXISTS ( SELECT 1 FROM pg_catalog.unnest(coalesce(pip.initprivs,pg_catalog.acldefault('c',c.relowner))) AS init(init_acl) WHERE acl = init_acl)) as foo) AS attacl, (SELECT pg_catalog.array_agg(acl ORDER BY row_n) FROM (SELECT acl, row_n FROM pg_catalog.unnest(coalesce(pip.initprivs,pg_catalog.acldefault('c',c.relowner))) WITH ORDINALITY AS initp(acl,row_n) WHERE NOT EXISTS ( SELECT 1 FROM pg_catalog.unnest(coalesce(at.attacl,pg_catalog.acldefault('c',c.relowner))) AS permp(orig_acl) WHERE acl = orig_acl)) as foo) AS rattacl, NULL AS initattacl, NULL AS initrattacl FROM pg_catalog.pg_attribute at JOIN pg_catalog.pg_class c ON (at.attrelid = c.oid) LEFT JOIN pg_catalog.pg_init_privs pip ON (at.attrelid = pip.objoid AND pip.classoid = 'pg_catalog.pg_class'::pg_catalog.regclass AND at.attnum = pip.objsubid) WHERE at.attrelid = '119531955'::pg_catalog.oid AND NOT at.attisdropped AND ((SELECT pg_catalog.array_agg(acl ORDER BY row_n) FROM (SELECT acl, row_n FROM pg_catalog.unnest(coalesce(at.attacl,pg_catalog.acldefault('c',c.relowner))) WITH ORDINALITY AS perm(acl,row_n) WHERE NOT EXISTS ( SELECT 1 FROM pg_catalog.unnest(coalesce(pip.initprivs,pg_catalog.acldefault('c',c.relowner))) AS init(init_acl) WHERE acl = init_acl)) as foo) IS NOT NULL OR (SELECT pg_catalog.array_agg(acl ORDER BY row_n) FROM (SELECT acl, row_n FROM pg_catalog.unnest(coalesce(pip.initprivs,pg_catalog.acldefault('c',c.relowner))) WITH ORDINALITY AS initp(acl,row_n) WHERE NOT EXISTS ( SELECT 1 FROM pg_catalog.unnest(coalesce(at.attacl,pg_catalog.acldefault('c',c.relowner))) AS permp(orig_acl) WHERE acl = orig_acl)) as foo) IS NOT NULL OR NULL IS NOT NULL OR NULL IS NOT NULL)ORDER BY at.attnum
pg_dumpall: error: pg_dump failed on database "ankara", exiting

```
When tested the failing query on a DB Client, it returns zero rows but without error. The postgresql log attached.


I then wrote a simple bash to `pg_dump` the databases one by one.  
```
#!/bin/bash

declare -a databases=('lugano' 'kinta' 'quito' 'zagreb')
for db in "${databases[@]}"; do
  echo "Starting pg_dump for ${db}."
  dt=`date +%Y%m%d-%T`
  cmd="pg_dump -Fc -b -v ${db} > /bnas/pgdumps/${db}_${dt}.tar"
  echo "${cmd}"
  eval "${cmd}"
  echo "pg_dump for ${db} completed."
done
```
Similarly, the first dump always succeeds, but the second one onwards fails. The postgresql log is also attached below. That said, triggering pg_dump manually one by one works, though painful.  

The server is a Dell T430 tower, with 12 core and 128 RAM. My `postresql.conf` is attached as well.  I suspect the checkpoint settings might have screw things up, cause I used to be able to continuously dump the databases one by one from bash script.