Prometheus times out when trying to scrape the table_metrics endpoint, for example, the below curl request can get stuck:
curl -H "Authorization: Bearer xxxxx" http://<FQDN>:6162/table_metrics
While at the same time, the endpoints for other metrics for example cluster_metrics can work fine
Greenplum Exporter work with Prometheus, here Greenplum Exporter is a comprehensive Prometheus exporter for Greenplum Database, providing detailed metrics for monitoring cluster health, performance, and resource utilization.
As advised in the documentation, table metrics come from the table gpmetrics.gpcc_table_info in gpperfmon database, we should consider whether the table is bloat or with missing statistics, which can result in notable slowness when querying it.
Run VACUUM/VACUUM FULL + ANALYZE against gpmetrics.gpcc_table_info
refer to KB 431358 for more information on another possible failure scenario.