Fix cert provisioner clearing valid challenges on transient errors
Signed-off-by: Bryan Frimin <bryan@getprobo.com>
This commit is contained in:
@@ -17,6 +17,7 @@ package certmanager
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
"github.com/getprobo/probo/pkg/coredata"
|
||||
@@ -115,6 +116,21 @@ func (p *Provisioner) checkPendingDomains(ctx context.Context) error {
|
||||
)
|
||||
}
|
||||
|
||||
func isFatalChallengeError(err error) bool {
|
||||
if err == nil {
|
||||
return false
|
||||
}
|
||||
|
||||
errStr := strings.ToLower(err.Error())
|
||||
|
||||
return (strings.Contains(errStr, "invalid") &&
|
||||
(strings.Contains(errStr, "challenge") ||
|
||||
strings.Contains(errStr, "authorization") ||
|
||||
strings.Contains(errStr, "order"))) ||
|
||||
strings.Contains(errStr, "authorization must be pending") ||
|
||||
strings.Contains(errStr, "expired")
|
||||
}
|
||||
|
||||
func (p *Provisioner) handleStaleProvisioningAttempts(ctx context.Context, conn pg.Conn) error {
|
||||
var domains coredata.CustomDomains
|
||||
if err := domains.ListStaleProvisioningDomains(ctx, conn, coredata.NewNoScope()); err != nil {
|
||||
@@ -291,35 +307,50 @@ func (p *Provisioner) provisionDomainCertificate(
|
||||
return nil
|
||||
}
|
||||
|
||||
if isFatalChallengeError(err) {
|
||||
p.logger.InfoCtx(
|
||||
ctx,
|
||||
"fatal challenge error, resetting domain to retry with fresh challenge",
|
||||
log.String("domain", domain.Domain),
|
||||
log.Int("retry_count", fullDomain.SSLRetryCount),
|
||||
)
|
||||
|
||||
fullDomain.HTTPChallengeToken = nil
|
||||
fullDomain.HTTPChallengeKeyAuth = nil
|
||||
fullDomain.HTTPChallengeURL = nil
|
||||
fullDomain.HTTPOrderURL = nil
|
||||
fullDomain.SSLStatus = coredata.CustomDomainSSLStatusPending
|
||||
|
||||
if updateErr := fullDomain.Update(ctx, conn, coredata.NewNoScope(), p.encryptionKey); updateErr != nil {
|
||||
p.logger.ErrorCtx(
|
||||
ctx,
|
||||
"cannot reset domain for retry",
|
||||
log.String("domain", domain.Domain),
|
||||
log.Error(updateErr),
|
||||
)
|
||||
return updateErr
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
p.logger.InfoCtx(
|
||||
ctx,
|
||||
"resetting domain to retry with fresh challenge",
|
||||
"transient error, keeping existing challenge for retry",
|
||||
log.String("domain", domain.Domain),
|
||||
log.Int("retry_count", fullDomain.SSLRetryCount),
|
||||
)
|
||||
|
||||
fullDomain.HTTPChallengeToken = nil
|
||||
fullDomain.HTTPChallengeKeyAuth = nil
|
||||
fullDomain.HTTPChallengeURL = nil
|
||||
fullDomain.HTTPOrderURL = nil
|
||||
fullDomain.SSLStatus = coredata.CustomDomainSSLStatusPending
|
||||
|
||||
if updateErr := fullDomain.Update(ctx, conn, coredata.NewNoScope(), p.encryptionKey); updateErr != nil {
|
||||
p.logger.ErrorCtx(
|
||||
ctx,
|
||||
"cannot reset domain for retry",
|
||||
"cannot update domain retry tracking",
|
||||
log.String("domain", domain.Domain),
|
||||
log.Error(updateErr),
|
||||
)
|
||||
return updateErr
|
||||
}
|
||||
|
||||
p.logger.InfoCtx(
|
||||
ctx,
|
||||
"domain reset to pending, will retry with new challenge on next cycle",
|
||||
log.String("domain", domain.Domain),
|
||||
)
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
|
||||
@@ -208,7 +208,7 @@ func (r *Renewer) renewDomain(ctx context.Context, conn pg.Conn, domain *coredat
|
||||
return updateErr
|
||||
}
|
||||
|
||||
return nil
|
||||
return fmt.Errorf("domain marked as failed after %d retry attempts: %w", maxRetries, err)
|
||||
}
|
||||
|
||||
// Update retry tracking but keep domain ACTIVE for next renewal cycle
|
||||
@@ -229,7 +229,8 @@ func (r *Renewer) renewDomain(ctx context.Context, conn pg.Conn, domain *coredat
|
||||
log.Int("retry_count", lockedDomain.SSLRetryCount),
|
||||
)
|
||||
|
||||
return nil
|
||||
// Return the original error so caller knows renewal failed
|
||||
return fmt.Errorf("renewal failed, will retry: %w", err)
|
||||
}
|
||||
|
||||
r.logger.InfoCtx(
|
||||
|
||||
Reference in New Issue
Block a user