PostgreSQL Autovacuum – Does Autovacuum Wake Up in Slave?

VACUUM processes launched by autovacuum can be safely killed with:

SELECT pg_terminate_backend(PID_of_backend);

Actually, any client processes in Postgresql can be terminated this way. Uncommitted work by this backend will be simply discarded.

You could then re-run VACUUM manually at a low-traffic time:

VACUUM VERBOSE users;

Check whether Cost-based Vacuum Delay can help you. This would limit the amount of I/O your autovacuum process uses.

Maybe you simply hit your IOPS limit. You should be able to see the numbers on AWS interface. On standalone Linux use iostat -dtkxy 10 to measure I/O. (iostat is usually packaged in sysstat package).

Maybe the VACUUM kicks in again so often because of the aggressive settings in your config.

Work done by an interrupted VACUUM FULL will be entirely lost, as it will simply revert to using the previous version of the table and throw away the in-progress version of the table.

Work done by a regular (not-FULL) VACUUM might not be entirely lost. It cleans indexes in batches, and any batches which were fully cleaned will not need to be cleaned again. They will still need to be inspected again, but will be found already clean the next time. So you might save some write IO which won't need to be repeated.