Claus Ibsen created CAMEL-26732:
-----------------------------------
Summary: camel-chaos - what to try, tried and test with AI: four
findings from a hands-on round
Key: CAMEL-26732
URL: https://issues.apache.org/jira/browse/CAMEL-26732
Project: Camel
Issue Type: Bug
Components: camel-jbang
Reporter: Claus Ibsen
Assignee: Claus Ibsen
Fix For: 4.24.0
Found by running an Orders demo (HTTP call, seda hand-offs, a dead letter
channel, a file poller) with camel run --chaos and trying what to try (w,
chaos-suggest, CAMEL-26727), what has been tried (CAMEL-26728) and test with AI
(t, CAMEL-26725) by hand:
# *The dead letter channel is suggested once per route.* With a context-wide
deadLetterChannel("seda:dlq"), what to try lists "The dead letter channel is
unreachable" for every route (6 times, the same rule endpoint=seda:dlq error
at=dlq), even for the route that consumes seda:dlq. It is one thing to try.
# *The hand-off suggestion is never marked tried.* The suggestion for route
orders (kind lost) is route=shipping error ... expect=orders->shipping:lost<=0.
A rule added from it fires in route shipping and is of kind consumer, so route
orders keeps "lost: not tried yet" in the w list and in the report.
# *camel cmd chaos-suggest does not print the tried mark* that the suggestion
has (tried=true), which the TUI w list shows.
# *The scenario that t writes does not replay the same faults.* A rule draws
from the seed by a key: its id (r2) when added live, its place in the file (r1)
in a scenario. So the scenario of rule r2, with the same seed, fails other
messages. Proposal: a rule option draw=<key> to draw as the rule with that id
did, which t writes.
Related to CAMEL-26607 (chaos stage 2).
_Filed with Claude Code on behalf of Claus Ibsen_
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
