novox/mesh-controller
6
0
Fork 0
Code Issues Pull Requests 5 Releases Wiki Activity

Send the bus's machine before the grants, and only when its user list moved (hq issue 249)

Browse Source 
Grants first could hold back the very declaration that lets the controller
issue them. The holder now goes first, then buckets and memberships, then
the rest; a membership that fails holds back only its own machine, and an
announcement whose send stopped at its grants is asked again. Whether the
holder must go first is read from a digest of the user list it was last
sent, not its whole declaration. Migration renumbered to 0058.
This commit is contained in:
jochen
2026-10-05 18:17:52 +02:00
parent 672d1f4ca1
commit ed90771382
9 changed files with 290 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files
cmd/mesh-controller/upgrades.go
+19 -5
View File
@@ -65,13 +65,17 @@ func (f following) Upgraded(ctx context.Context, u link.Upgraded) error {
if plans, err := inv.OpenPlans(ctx); err != nil {
return notNow(err)
} else if id := rolledOutByAPlan(plans, u.Module); id != "" {
fmt.Printf("%s moved to %s; %s rolls it out to %s\n", u.Module, shortCommit(u.Commit), id, readableList(on))
// Said with its remedy: a plan that ends without sending it — failed, or closed by hand — leaves
// these machines behind, which `status` lists and `push --behind` sends (novox/hq issue 249).
fmt.Printf("%s moved to %s; %s rolls it out to %s — if that plan ends without sending it, "+
"`status` lists them as behind and `push --behind` sends it\n",
u.Module, shortCommit(u.Commit), id, readableList(on))
return nil
}
if decision.Together {
fmt.Printf("%s moved to %s; sending %s together\n",
u.Module, shortCommit(u.Commit), readableList(on))
return sendTo(ctx, f.open, on)
return askAgainOnGrants(sendTo(ctx, f.open, on))
}
// One at a time, and stopping at the first that fails.
//
@@ -82,18 +86,28 @@ func (f following) Upgraded(ctx context.Context, u link.Upgraded) error {
u.Module, shortCommit(u.Commit), readableList(on))
for _, node := range on {
if err := sendTo(ctx, f.open, []string{node}); err != nil {
return fmt.Errorf("%s did not take %s, so the machines after it were left alone: %w",
node, u.Module, err)
return askAgainOnGrants(fmt.Errorf("%s did not take %s, so the machines after it were left alone: %w",
node, u.Module, err))
}
}
return nil
}
// askAgainOnGrants marks a send that stopped at its grants as one to ask again (novox/hq issue 249):
// an announcement handled by a send whose memberships could not be issued is held and redelivered,
// rather than taken as handled with the machines left on the old version.
func askAgainOnGrants(err error) error {
if err != nil && errors.Is(err, errGrants) && !errors.Is(err, link.ErrTryAgain) {
return fmt.Errorf("%w: %w", link.ErrTryAgain, err)
}
return err
}
// rolledOutByAPlan is the open plan that will send a module's machines its new build — one holding
// the module that has not finished sending it — or empty when none will (novox/hq issue 249).
func rolledOutByAPlan(plans []inventory.Plan, module string) string {
for _, p := range plans {
if s, holds := p.Modules[module]; p.Open() && holds && (s == nil || s.SentAt == nil) {
if s, holds := p.Modules[module]; p.Open() && holds && (s == nil || (s.SentAt == nil && s.State != "failed")) {
return p.ID
}
}