Job logs and metrics

Two Job verbs read what the platform’s monitoring service recorded while a Job ran. Both take the Job’s UUID or URL and connect like every other resource command.

Logs

job logs <ref> prints the Job’s log entries, oldest first, one per line:

pinexq job logs 3f2b…
pinexq job logs 3f2b… --level warn
pinexq job logs 3f2b… --search "timeout" --tail 50

Filters:

• --level LEVEL keeps that level and above. Levels, ascending: TRACE, DEBUG, INFO, WARN, ERROR, FATAL, CRITICAL. An entry with an unknown level always passes.
• --search TEXT keeps entries whose message contains the text (case-insensitive).
• --tail N prints only the last N matching entries.

Output formats (-o, --output):

• text (default) — [timestamp] LEVEL message, coloured by level on a terminal. --no-color or NO_COLOR turns colour off.
• raw — the message only.
• json — one JSON object per line (NDJSON, not an array), with the timestamp, level, message, and the entry’s metadata.

Note

The filters run in the CLI, so job logs fetches every page of the Job’s logs unless you use --tail. Line-oriented output means grep, wc -l, and per-line jq work in every format.

pinexq job logs 3f2b… -o json | jq -r 'select(.level == "ERROR") | .message'

Metrics

job metrics <ref> shows the CPU and memory series sampled from the worker that ran the Job:

pinexq job metrics 3f2b…
pinexq job metrics 3f2b… -o csv > metrics.csv

• text (default) — a header with the peak CPU and memory over the Job’s lifetime.
• json — one object with the full series: usage in cores and bytes, and utilization as a percentage of the worker’s requested resources.
• csv — one row per sample timestamp.

Metrics exist only for Jobs the platform ran after the monitoring link was introduced; older Jobs report that no metrics are available.