The following bug has been logged on the website:
Bug reference: 16595
Logged by: Jesse Kinkead
Email address: jesse@falkon.ai
PostgreSQL version: 11.6
Operating system: Linux
Description:
We've seen this on a few different servers running Postgres 11.6.
The table in question has a JSONB column with a GIN index, defined as:
"index_name_idx" gin (jsonb_column_name jsonb_path_ops, integer_column,
timestamp_with_time_zone_column)
Under load (especially writes followed immediately by reads), we
occasionally see reads failing an internal exception (error code XX000) with
the message "lost saved point in index".
This seems to come from this source file:
https://doxygen.postgresql.org/ginget_8c_source.html
To further complicate matters, this server is using TimescaleDB
(https://www.timescale.com/), and has this table configured as a hypertable
(https://docs.timescale.com/latest/using-timescaledb/hypertables), which I
believe means it's actually multiple tables masquerading as a single
table.
The errors are quite frequent when running with slow CPUs / disks, but more
rare when running with more resources.