Demo: 30 days of history so the graphs aren't empty #39

Closed
opened 2026-11-16 05:21:17 +00:00 by jcoffey-dev · 1 comment
Owner

Every Day / 30 days / Year graph on the demo says "hourly averages: none kept yet", because the container starts fresh each week. A monitoring demo with no trends is missing its main point, and the docs screenshots come from it.

Do: the fleet sim writes believable hourly averages for the last 30 days (daily cycle, a slow disk growth, one past incident) into the store before obs-beacond starts, and a matching set of resolved incidents.

Every Day / 30 days / Year graph on the demo says "hourly averages: none kept yet", because the container starts fresh each week. A monitoring demo with no trends is missing its main point, and the docs screenshots come from it. **Do:** the fleet sim writes believable hourly averages for the last 30 days (daily cycle, a slow disk growth, one past incident) into the store before obs-beacond starts, and a matching set of resolved incidents.
jcoffey-dev added the
priority
p2
enhancement
labels 2026-11-16 05:21:17 +00:00
jcoffey-dev self-assigned this 2026-11-16 05:21:17 +00:00
Author
Owner

Done in #51. fleet run -history writes the last 30 days into obs-beacond's database before anything reports, from the same plan and clock the live demo runs on: every host's hours (a working day per site time zone, quieter weekends, memory that creeps until a weekly restart, disks that fill until a clean-up), cluster nodes, array I/O, IBM i CPU and Vault leases for the month, and a week of controllers and VMs, which keeps the database at about 90 MB. It also writes mail rounds, availability for the reports, and an incident for each past spell, with notes on the stories. Incident numbers carry on from there. The demo container's entrypoint gets the flag with the next demo deploy.

Done in #51. `fleet run -history` writes the last 30 days into obs-beacond's database before anything reports, from the same plan and clock the live demo runs on: every host's hours (a working day per site time zone, quieter weekends, memory that creeps until a weekly restart, disks that fill until a clean-up), cluster nodes, array I/O, IBM i CPU and Vault leases for the month, and a week of controllers and VMs, which keeps the database at about 90 MB. It also writes mail rounds, availability for the reports, and an incident for each past spell, with notes on the stories. Incident numbers carry on from there. The demo container's entrypoint gets the flag with the next demo deploy.
Sign in to join this conversation.