Xray
hourly rollups 1h 6h 24h 48h 3d 5d 7d 14d 30d ‹
Last 48 hours▾
Ends now. Anything up to this cluster's 30-day retention.
Start and end times
Cancel
Interpreted in UTC (change it at the right of this bar). Earliest available: 2026-10-19 12:33:16.
› →|
updated 1d ago
Sample data. This database was generated by scripts/seed_dev_db.py — every service, deploy, person and figure on this page is synthetic. Names are sample.*@seed.invalid and image tags are prefixed SEED-. Nothing here happened.
← findings

chat-agent: p99 latency (ms) 42.90 s — 12.5× the fleet's usual 3.42 s (also p95 latency (ms) 9.61 s)

info Service outlier test27 open impact 10 Owner: Sambasiva Bandarupalli (backup: Ahmed Aburaed) opened 76d ago · last observed 1d ago · observed 764× · #697 · 278c0211ffcad112792e

What happened template

chat-agent differs from the other 8 services on 2 measures over six hours: p99 latency (ms) 42.90 s against the peers' 3.42 s (15.8 MADs above); p95 latency (ms) 9.61 s against the peers' 689 ms (9.8 MADs above). Peer comparison is structural more often than it is news — a service that does more work per request will sit here for as long as it does — which is why this is informational unless it is about failures.

A service unlike the rest of the fleet on one measure (robust z against its peers).

The facts

measures{'measure': 'p99 latency (ms)', 'value': 42899.99999999999, 'value_fmt': '42.90 s', 'median': 3418.440677966103, 'median_fmt': '3.42 s', 'mad': 2492.102340000011, 'z': 15.842671742779922, 'direction': 'above', 'peers': 8}, {'measure': 'p95 latency (ms)', 'value': 9614.583333333332, 'value_fmt': '9.61 s', 'median': 689.2721718980408, 'median_fmt': '689 ms', 'mad': 914.3379702252743, 'z': 9.761501164866068, 'direction': 'above', 'peers': 8}
measurep99 latency (ms)
value42,900
value_fmt42.90 s
median3,418
median_fmt3.42 s
z15.84
directionabove
peers8
rule{"min_score": 3.5, "min_relative_pct": 35, "warn_on": []}
discriminator
window2026-11-16 15:01:22 UTC → 2026-11-16 21:01:22 UTC

What is failing in chat-agent — the bad events, decomposed

“Bad events” are HTTP 5xx answers from chat-agent. In this finding's window (2026-11-16 15:01:22 UTC → 2026-11-16 21:01:22 UTC) there were 0; every line links to the grouped errors behind it.

The error shapes behind them

10× hitl_tool: pre-approval card generation failed since 2026-11-15 15:19:25 UTC
5× status_tool: body raised, surfaced to the LLM as a tool error since 2026-11-16 17:58:36 UTC

All chat-agent errors, grouped → · service page · its logs

History

  • 2026-11-15 18:56:44 UTCreopened x-ray · severity=info was=info title=chat-agent: p95 latency (ms) 14.00 s — 28.8× the fleet's usual 486 ms (also p99 latency (ms) 24.40 s) level=low was_level=low finding={'id': 697, 'fingerprint': '278c0211ffcad112792e', 'kind': 'outlier.service', 'severity': 'info', 'cluster': 'test27', 'subject': 'chat-agent', 'title': "chat-agent: p95 latency (ms) 14.00 s — 28.8× the fleet's usual 486 ms (also p99 latency (ms) 24.40 s)", 'facts': {'measures': [{'measure': 'p95 latency (ms)', 'value': 13999.999999999985, 'value_fmt': '14.00 s', 'median': 485.6587837837838, 'median_fmt': '486 ms', 'mad': 607.2542128378378, 'z': 22.254833199197737, 'direction': 'above', 'peers': 5}, {'measure': 'p99 latency (ms)', 'value': 24399.999999999978, 'value_fmt': '24.40 s', 'median': 1146.249999999995, 'median_fmt': '1.15 s', 'mad': 1647.8661365060166, 'z': 14.111431435386548, 'direction': 'above', 'peers': 5}], 'measure': 'p95 latency (ms)', 'value': 13999.999999999985, 'value_fmt': '14.00 s', 'median': 485.6587837837838, 'median_fmt': '486 ms', 'z': 22.254833199197737, 'direction': 'above', 'peers': 5, 'window': [1794747404, 1794769004], 'recent_deploys': [], 'rule': {'min_score': 3.5, 'min_relative_pct': 35, 'warn_on': []}, 'discriminator': ''}, 'evidence': [{'label': 'Outliers', 'href': '/outliers?cluster=test27&from=1794747404&to=1794769004'}, {'label': 'chat-agent service page', 'href': '/service/chat-agent?cluster=test27&from=1794747404&to=1794769004'}, {'label': 'Tune this rule', 'href': '/alerts/system/outliers'}], 'provenance': {'source': 'rollup', 'collected_at': 1794769002, 'age_seconds': 3, 'stale': False, 'shown': None, 'total': None, 'truncated': False, 'coverage_from': None, 'retain_days': None, 'note': None}, 'impact': 5.1, 'status': 'open', 'first_seen': 1788438316, 'last_seen': 1794769004, 'resolved_at': None, 'acked_by': None, 'acked_at': None, 'snoozed_until': None, 'occurrences': 697, 'narrative': "chat-agent differs from the other 7 services on 2 measures over six hours: p99 latency (ms) 17.25 s against the peers' 4.53 s (24.8 MADs above); p95 latency (ms) 8.12 s against the peers' 484 ms (11.1 MADs above). Peer comparison is structural more often than it is news — a service that does more work per request will sit here for as long as it does — which is why this is informational unless it is about failures.", 'narrative_kind': 'template', 'discriminator': '', 'level': 'low', 'told_level': None, 'episode_at': 1794769004, 'chat_thread': None, 'kind_label': 'Service outlier', 'group': 'Anomalies', 'age_s': 6330688, 'stale_s': 0}

Check it yourself

Every link carries the finding's own time window, so it shows tomorrow what it shows now.

Actions

Sign in (IAP) to acknowledge, snooze or resolve.

Where the numbers come from

sourcerollup
collected_at1794862880
age_seconds3
staleFalse
shownNone
totalNone
truncatedFalse
coverage_fromNone
retain_daysNone
noteNone
detectoroutliers