Repository navigation
PR watch idle since 2026-10-28 15:06 UTC: ready fcvm-fleet PRs #714, #717, #720 not taken #721
Description
The jumpbox's pr-watch has merged nothing since #719 (2026-10-28 15:06 UTC). Three ready ejc3-authored PRs are waiting:
- fcvm-fleet: the wall's broadcast on Amazon IVS #714 fcvm-fleet: the wall's broadcast on Amazon IVS (ready since 2026-10-29 05:20 UTC; its ejc3/fcvm side is deployed: coordinator image from main 407f4e16 and later, page from 38fd9a9e). Its one Codex finding is answered (the fleet test suite updated, 53 passed).
- fcvm-fleet: tile AMI from ejc3/fcvm main 71bb448e (boot prefetch) #717 fcvm-fleet: tile AMI from ejc3/fcvm main 71bb448e.
- fcvm-fleet: tile root volumes at gp3 1000 MiB/s and 4000 IOPS #720 fcvm-fleet: tile root volumes at gp3 1000 MiB/s and 4000 IOPS.
Asking the jumpbox to take its next run; #714 blocks the broadcast's staging verification.
Activity
ejc3 commentedon Oct 29, 2026
Owner-triggered review round completed: #717 and #720 are merged and applied at main 7cf2ccd. Combined 52 ARM tests passed; AWS readback verifies template versions 8/8/5, the new private AMI and gp3 settings. Coordinator revision 55 remains healthy, and the final full Terraform plan is empty.
#714 is not safe to enable yet. Its IAM inline-quota overflow and first-deployment secret/grant ordering were corrected in 67742d0 (format follow-up 156ae9a). One genuine runtime-image dependency remains: current host/tile binaries cannot enforce the new admission keys/direct-viewer limit. Exact source evidence is in its unresolved review thread. New admission-capable host and tile runtime artifacts are required before merge.
This was a manual owner-triggered round; it does not establish that the recurring watch’s existing scheduling/quota problem is repaired. Keeping this issue open rather than claiming all three are complete.