After setting the option "num-write-checks-before-scale-down" to 3, the GSI is never updated because the consecutive write checks appears to be stuck at the value 1:
INFO - scorehistory - GSI: scoreProfileID-timeQuantum-index - Consecutive write checks 1/3
The above is logged in every check interval. On code inspection, found what appears to be a cut-paste error in init.py. In the below code, CHECK_STATUS['tables'][table_name]['reads'] should be replaced by CHECK_STATUS['gsis'][gsi_name]['reads']. Similar for writes.
for gsi_name, gsi_key in sorted(gsi_names):
gsi_num_consec_read_checks = \
gsi_num_consec_read_checks = 0
gsi_num_consec_write_checks = \
gsi_num_consec_write_checks = 0
Thank you for the report. I'll look into this and I aim to release a new version today.
This is blocked by #207 at the moment. #207 is waiting for AWS to roll out their new release to all regions, which should be done early next week.
OK. For the record, I patched 1.18.5 with the change indicated in the issue description above and it fixed the problem with GSI's.
Super, thanks for the info!
Fixed consecutive check issue for GSIs #206