System Troubleshooting

Application logs can be obtained from journald to report and investigate issues. The journald documentation provides advanced usage information and information on log archival or sending logs to remote log processing frameworks.

A few example commands for journald logs are given here.

journalctl -u hpcd
journalctl -u -xe hpcd
journalctl -u hpcd --since "2024-12-24 23:59:00"
journalctl -u hpcd --since "3 hours ago" --until "1 hour ago"

For any suspected performance issue, information can be gathered during the issue (i.e., generally at peak, while CPU is overloaded) through the following relatively non-intrusive command.

sudo perf script flamegraph -ag -F99 --call-graph lbr -- sleep 10

🚧

IMPORTANT

EdgePeak is a sharded server (i.e., one worker per core, one thread per core), which means that some internal structures such as connection pools are maintained per-core.