10.5 but not 10.4: backend startup during reindex system: could notread block 0 in file "base/16400/..": read only 0 of 8192 bytes - Mailing list pgsql-hackers

I've seen this message now a handful of times recently.  It seems to happen
overnight, during a maintenance job which reindex things, including system
catalog indices.

It's easy to reproduce error under 10.5, but not under 10.3 nor 10.4.

while :; do for a in pg_class_oid_index pg_class_relname_nsp_index pg_class_tblspc_relfilenode_index; do psql ts -qc
"REINDEXINDEX $a"; done; done&
 

[pryzbyj@database ~]$ while :; do psql ts -qc ''; done
psql: FATAL:  could not read block 1 in file "base/16400/313430687": read only 0 of 8192 bytes
psql: FATAL:  could not read block 0 in file "base/16400/313430708": read only 0 of 8192 bytes
psql: FATAL:  could not read block 0 in file "base/16400/313430711": read only 0 of 8192 bytes
^C

postgres=# SELECT * FROM postgres_log WHERE error_severity ='FATAL';
log_time               | 2018-08-28 22:19:58.822-05
user_name              | telsasoft
database               | ts
pid                    | 22445
connection_from        | 192.168.122.12:58318
session_id             | 5b8610de.57ad
session_line           | 1
command_tag            | startup
session_start_time     | 2018-08-28 22:19:58-05
virtual_transaction_id | 26/280967
transaction_id         | 0
error_severity         | FATAL
sql_state_code         | XX001
message                | could not read block 0 in file "base/16400/313316300": read only 0 of 8192 bytes

Justin