Documentation
¶
Overview ¶
Package replay re-applies the host-side records in $OBOL_CONFIG_DIR to the cluster. Every piece of operator state that only lives in etcd has a record-on-write file next to the stack config; ReplayRecorded is the ONE ordered list that brings a fresh (or reset) cluster back in line with them. `obol stack up` and `obol sell resume` both call it, so a host reboot and a full cluster recreation converge on the same state.
The order is defined, and justified step by step, on Order.
Every step is best-effort: a failure (returned error or any warning the step printed) is reported per step and in the closing summary, and never aborts the remaining steps or `stack up`.
Index ¶
Constants ¶
const ( StepModels = "models" StepNetworks = "networks" StepRPCUpstreams = "rpc-upstreams" StepERPCOverlay = "erpc-overlay" StepX402Pricing = "x402-pricing" StepAgentIdentity = "agent-identity" StepAgentInstances = "agent-instances" StepAgentCRs = "agent-crs" StepStorefront = "storefront" StepApps = "apps" StepSellOffers = "sell-offers" )
Step names, in replay order (see the package doc for why).
Variables ¶
var Order = []string{ StepModels, StepNetworks, StepRPCUpstreams, StepERPCOverlay, StepX402Pricing, StepAgentIdentity, StepAgentInstances, StepAgentCRs, StepStorefront, StepApps, StepSellOffers, }
Order is the single replay list shared by `obol stack up` and `obol sell resume`. Each step depends only on steps above it.
Functions ¶
This section is empty.
Types ¶
type Options ¶
type Options struct {
// ResumeSellOffers replays persisted ServiceOffers and warns about
// legacy cluster-routed sell-inference descriptors (cmd/obol resumeSellOffers). nil skips
// the step.
ResumeSellOffers func(ctx context.Context, cfg *config.Config, u *ui.UI) error
}
Options carries the steps implemented outside internal/ (package main).