Distribution error codes
Structured per-target error codes emitted by every distribution module, the error class each code maps to, and which classes the fan-out per-target retry re-attempts. Codes appear in the error_code field of the per-target results (last_target_results) on the distribution and in job logs.
Error classes
Retry policy is based on a normalized error class, not on raw error strings. Each code maps to one of five classes; only the first two are re-attempted by fan-out per-target retry.
| Class | Meaning | Retried by fan-out? |
|---|---|---|
network | Timeout, connect failure, DNS resolution failure. | Yes |
io_transient | Broken pipe, connection reset, transient SFTP/API failure — likely to succeed on retry. | Yes |
io_permanent | Remote permission denied, no space left, quota/ownership conflicts — operator action needed. | No |
auth | SSH/API authentication failure, credential rejected. | No |
validation | Cert/key mismatch, invalid path, missing target — the request itself is wrong. | No |
A code that is not mapped to any class (and where the module did not set a class itself) is treated as not retryable.
Cross-module codes
| Code | Class | Meaning |
|---|---|---|
CREDENTIAL_MISSING | auth | The module credential referenced by the target could not be resolved. |
TLS_TRUST_REFUSED | validation | PAN-OS, F5 BIG-IP, NetScaler, FortiWeb, FortiGate, A10 Thunder: the target's own TLS trust options refused the connection — the appliance presented a certificate other than the pinned tls_fingerprint_sha256, or the options (ca_cert, the pin) could not be applied. Nothing was sent, so it is not retried and the healthy targets of the same run are not rolled back for it. The error names the fingerprint the appliance presented; update the pin when the management certificate changed, or clear it. (Before 1.7.0 a pin mismatch surfaced as the module's *_CONNECT code and was retried.) |
Credentials are bound to the hosts they are used with (which hosts a credential may be sent to). A deploy, dry run or health check whose target would send a credential to a host it is not bound to stops before the secret is read, at the job level: the job's last error starts with CREDENTIAL_HOST_NOT_ALLOWED and names the credential and the host. Saving such a pairing is refused the same way (422).
The same rule answers 422 on the API before anything is written: CREDENTIAL_HOST_NOT_ALLOWED (a target, group, override, store or live picker would send a bound credential to a new host), CREDENTIAL_TLS_CHANGE_NOT_ALLOWED (turning down how a bound target checks the server it sends the secret to — ssl_verify / use_tls off, tls_skip_verify on; on a PAN-OS / F5 / NetScaler / FortiWeb / FortiGate target also turning ssl_verify off even while adding a tls_fingerprint_sha256 pin, adding a ca_cert beside the system roots while ssl_verify is on, or replacing or removing a pin or a ca_cert; adding a pin while ssl_verify stays on, or adding a pin or CA to a target that verified nothing, is free; a new target that only pins is judged like one that does not verify), CREDENTIAL_REF_CHANGE_NOT_ALLOWED (pointing a store-backed credential at another secret_ref), CREDENTIAL_SECRET_REQUIRED (moving the KDC / domain inside a WinRM-type secret without the password), DNS_CREDENTIAL_SECRET_REQUIRED and MSCA_CONNECTION_CREDENTIAL_REQUIRED (moving where a DNS credential or AD CS connection connects without its secret). Each message names who may do it: the person who entered the secret, or an organization owner.
Saving a distribution's overrides answers 422 OVERRIDE_GROUP_CONFLICT when two group rows that both set a value share a target, and 409 CLUSTER_UPGRADE_REQUIRED for any group row while a node of an older version is running.
A credential read from an external secret store is resolved before any target is attempted, so a store failure has no target to attach a code to. It is reported at the job level instead — the job status and last error name the credential, the store and the underlying failure. There is deliberately no SECRETSTORE_* code, because no target result could ever carry one.
SSH
| Code | Class | Meaning |
|---|---|---|
SSH_CONNECT | network | TCP/SSH connection to the target failed, or the target ran out of its time budget before the deploy finished (a slow or hung host). A job the operator cancelled carries no code and is not retried. |
SSH_AUTH | auth | The host rejected the credentials during the SSH handshake (key or password). A handshake that fails for a transport reason — the peer closes the connection, connection reset — is SSH_CONNECT, not this. |
SFTP_FAILED | io_transient | SFTP session could not be opened or an SFTP operation failed. |
FILE_UPLOAD | io_transient | Writing a PathSet file on the target failed for a transient reason (I/O error, lost connection, full disk). |
SSH_PERMISSION | io_permanent | The host refused the file operation: permission denied on the file or directory, a missing parent directory the SSH user could not create, a path that is not a directory, or a read-only filesystem (from the SFTP status code or its text). Not retried — give the SSH user write access, create the directory, or fix the PathSet path. |
SCRIPT_EXIT_NONZERO | io_transient | The ActionSet script ran but exited with a non-zero status. |
SCRIPT_FAILED | io_transient | Reserved. The SSH module reports a failing ActionSet as SCRIPT_EXIT_NON_ZERO; this code is not emitted today. |
SSH_VALIDATION | validation | Pre-flight validation of the SSH deployment inputs failed: the target has no SSH spec or its bound PathSet/ActionSet no longer resolves (deleted, another project, kind changed), a path could not be rendered, an unresolved or secret variable in a path, path traversal, an invalid owner or run_as, a pfx passphrase variable that is missing, not secret or empty. Also the fallback for any unclassified error. |
SSH_HOST_KEY_ALGO | validation | Host-key negotiation failed before any key was presented: the server offers only key types CertAutoPilot does not advertise once verification is enabled — in practice a DSA-only server. Not retried, and there is nothing to approve. Add a modern host key to the server, or turn verification off for that target. A server that merely dropped the pinned key type does not land here: the other types stay on offer, so it is reported as SSH_HOST_KEY_CHANGED and can be reviewed. |
SSH_HOST_KEY_CHANGED | validation | The server presented a host key that does not match the one pinned for this target. Nothing was deployed, and the run is not retried — review and approve the new key on the target. Only raised when host-key verification is enabled on that target. |
Webhook
| Code | Class | Meaning |
|---|---|---|
WEBHOOK_CONNECT | network | The webhook endpoint could not be reached on the last attempt (DNS, refused, timeout, TLS, policy refusal, redirect limit, unparseable URL). Connection failures are retried within retry_count; an unparseable URL is not. |
WEBHOOK_NON_2XX | io_transient | The endpoint answered with a non-2xx HTTP status on the last attempt. 5xx, 408, 425 and 429 are retried within retry_count; any other 4xx ends the attempts immediately. |
WEBHOOK_TEMPLATE | validation | The payload template failed to render (or the default payload could not be built). No request was sent and nothing is retried — fix the module-config template and run a dry run. |
Kubernetes
| Code | Class | Meaning |
|---|---|---|
K8S_CONNECT | network | The API server could not be reached (dial, DNS, TLS, timeout, 503/504), or a live-picker listing (namespaces / Secrets / workloads) failed for a reason other than RBAC. |
K8S_FORBIDDEN | auth | The API server answered 401 (expired or revoked token) or 403 (RBAC) for the secret create/update or workload patch — or, in a live picker, the list verb the listing needs. |
K8S_NOT_FOUND | validation | Namespace or referenced object does not exist. |
K8S_APPLY_FAILED | io_transient | Applying the Secret or the restart patch failed for another reason (conflict, immutable/wrong-type Secret, admission or quota rejection). |
K8S_VALIDATION | validation | The API client could not be built — kubeconfig credential missing, empty or unparseable, unknown context, in_cluster outside a pod, invalid proxy_url. Nothing was dialed; not retried. With Find by certificate names, also: the certificate carries no DNS name to match. |
K8S_SCOPE_TOO_LARGE | validation | Find by certificate names found more than 5000 TLS Secrets in the namespaces it searches. Nothing is written; name the namespaces to search. Not retried. |
Warnings a Find by certificate names run can add (the run still succeeds): K8S_NOTHING_MATCHED (no Secret holds the certificate's names — nothing written), K8S_SECRETS_SKIPPED (Secrets sharing a name were left alone; the job log names each and why), K8S_OWNED_UPDATED (a cert-manager / Helm Secret was updated because include owned is on), K8S_ROLLOUT_SKIPPED (a chosen workload did not roll: paused, OnDelete, it uses no updated Secret, or the credential cannot list a namespace's workloads), K8S_NAMESPACE_FORBIDDEN (a namespace in scope could not be read).
NetScaler
| Code | Class | Meaning |
|---|---|---|
NETSCALER_CONNECT | network | NITRO API unreachable at any step (refused, DNS, TLS, timeout, dropped connection), or the pre-flight nsversion probe did not answer like a NITRO endpoint. |
NETSCALER_AUTH | auth | HTTP 401/403 or NITRO 354 at any step — wrong password, or a command the user's command policy does not allow. |
NETSCALER_UPLOAD | io_transient | NetScaler rejected a cert/key systemfile write (rights, cert_path, disk space). |
NETSCALER_UPDATE | io_transient | NetScaler rejected the certkey create/update, a create conflict could not be resolved safely, or a certkey read failed for a non-transport, non-auth reason. |
NETSCALER_CHAIN | io_transient | Installing the intermediate certkeys failed (parse error, >5 intermediates, rejected upload/create, unresolvable content conflict). Strict mode only. |
NETSCALER_LINK | io_transient | The link step failed (link read, unlink, link, or the read-back showing a different link). Strict mode only, except a broken link that could not be restored, which fails in every mode. |
NETSCALER_OWNERSHIP | io_permanent | The object exists but is not managed by CertAutoPilot — refusing to overwrite. |
An appliance that presents a certificate other than the target's pinned tls_fingerprint_sha256 is the cross-module TLS_TRUST_REFUSED (nothing sent, not retried), not NETSCALER_CONNECT.
Huawei Cloud
| Code | Class | Meaning |
|---|---|---|
HUAWEICLOUD_CONNECT | network | Huawei Cloud API unreachable, HTTP 5xx, client initialisation failed, or a 2xx create response without an id. |
HUAWEICLOUD_AUTH | auth | AK/SK authentication rejected (401/403). |
HUAWEICLOUD_NOT_FOUND | validation | The referenced certificate object or CDN domain does not exist. |
HUAWEICLOUD_CONFLICT | io_permanent | Name/state conflict — e.g. ambiguous name lookup matched multiple certificates, or the name listing was cut off (WAF page incomplete, ELB more than 500 matches) with no exact match, so nothing is created. |
HUAWEICLOUD_QUOTA | io_transient | API quota / rate limit hit (429). |
HUAWEICLOUD_VALIDATION | validation | HTTP 400 — the API rejected the request content (invalid PEM, name or field). Not retried. |
F5 BIG-IP
| Code | Class | Meaning |
|---|---|---|
F5BIGIP_CONNECT | network | The device did not answer (unreachable, TLS/verify error, timeout, transport error on any call), or the first call failed with a non-auth HTTP error. A rejected login is F5BIGIP_AUTH. |
F5BIGIP_AUTH | auth | HTTP 401/403 anywhere in the run: a rejected token login (wrong password or login provider), a role denial on the upload endpoint, or a forbidden object/profile/transaction call. |
F5BIGIP_UPLOAD | io_transient | Uploading a cert/key file failed (5xx or transport error). A 401/403 role denial is F5BIGIP_AUTH. |
F5BIGIP_UPDATE | io_transient | TMOS below 13.0 or an unparseable version; creating or updating the cert/key objects failed; a transaction could not start, commit or timed out; the versioned name is too long. |
F5BIGIP_NOT_FOUND | validation | An import-only in-place update targeted an object that does not exist (F5 answered 404) — typically an explicit key_name whose ssl-key object was never created while the cert object exists. |
F5BIGIP_CHAIN | io_transient | Installing the intermediates failed (strict mode): chain unparseable, more than 5 intermediates, a lookup/upload/create failed, or an existing object conflicts with no identical object to reuse. |
F5BIGIP_OWNERSHIP | io_permanent | A cap_f5ic_<fp24> intermediate object exists but holds a different certificate (strict mode). |
F5BIGIP_PROFILE | io_permanent | Profile-aware only — the module never creates profiles: no listed client-ssl profile exists, a profile could not be read, its certKeyChain PATCH or commit failed or timed out, or it already serves a different certificate of the same key type. |
A BIG-IP that presents a certificate other than the target's pinned tls_fingerprint_sha256 is the cross-module TLS_TRUST_REFUSED (nothing sent, not retried), not F5BIGIP_CONNECT.
HashiCorp Vault
| Code | Class | Meaning |
|---|---|---|
VAULT_CONNECT_FAILED | network | Vault unreachable. |
VAULT_AUTH_FAILED | auth | AppRole login rejected, or Vault answered 403 on the secret read or write (invalid/expired token, or the policy lacks read / create / update on the path). Not retried. |
VAULT_READ_FAILED | io_transient | Reading the existing secret failed for a reason other than 403 (network, 5xx, unparsable body). Nothing is written. |
VAULT_WRITE_FAILED | io_transient | Writing the secret failed for a reason other than 403 (network, 5xx). |
VAULT_VALIDATE_FAILED | validation | Reserved: defined but not currently emitted. |
VAULT_PATH_INVALID | validation | The configured KV path is invalid. |
Windows (generic WinRM)
| Code | Class | Meaning |
|---|---|---|
WINRM_CONNECT | network | WinRM endpoint unreachable. |
WINRM_AUTH | auth | The Windows host refused the request at the HTTP layer (401/403 — or HTTP 500 "Access is denied" when the account may not open the WinRS shell, e.g. it is only in Remote Management Users), or its listener's TLS certificate is not trusted or was issued for another name. The message quotes the auth schemes the listener offered — a missing Negotiate means NTLM is disabled or blocked by policy; on plain HTTP a missing Basic challenge — or an HTTP 500 with an empty body — means AllowUnencrypted is false. Otherwise the credential or the account's WinRM authorization was refused (walkthrough). For Kerberos, check the realm, that the target is addressed by AD FQDN, and that this host's DNS resolves AD records (or set the KDC Address field). |
WINRM_TIMEOUT | io_transient | Command execution timed out. |
WINRM_EXEC | io_transient | ActionSet command/script failed on the host. |
WINRM_PS_SYNTAX | validation | The PowerShell payload has a syntax error, or calls a cmdlet that is not recognized (detected in the failed ActionSet output). |
WINRM_FILE | io_transient | The file transfer to the host failed, or what arrived did not match what was sent (the content check runs on the host and the destination is left untouched on a mismatch). |
WINRM_VALIDATION | validation | Operator configuration error: unresolved PathSet/ActionSet, invalid file path or missing variable, missing/non-secret pfx passphrase variable, invalid or absent shell, a command rejected by allowed_commands, or an unusable allowed_commands pattern. Nothing is retried. |
WINRM_QUOTA | io_permanent | WinRM quota exceeded (envelope size, concurrent shells) — including a transfer the host refused for its SIZE (HTTP 413): MaxEnvelopeSizekb is below what it needs, not the credential. A 401 during a run is reported as an authentication failure (the credential stopped working mid-run), with the size limit mentioned as the thing to check if it only ever happens on the file transfer. |
WINRM_HOST_PREREQ | validation | The host cannot run the module's scripts as it is: PowerShell in Constrained Language mode (health check and deploy), or a PowerShell older than 3, which cannot run the file transfer (health check). Reported by the health-check preflight or at deploy time (a non-administrator or UAC-filtered token, an unwritable %TEMP%, a low envelope limit and an old PowerShell are only warnings for this module); the message names the fix (catalogue). |
WINRM_OUTPUT_LIMIT | io_permanent | A command printed more than CertAutoPilot keeps of a command's output (64 MiB on standard output) and was stopped on the host — never an ActionSet command or script, which run to their end however much they print. Used by the Windows modules (WinRM, IIS, Exchange) and Windows DNS. Not retried: the same command prints the same again. A server certificate scan is not failed by it — the target is reported Incomplete with what fitted. |
Microsoft IIS
| Code | Class | Meaning |
|---|---|---|
IIS_CONNECT | network | WinRM connection or in-flight transport to the Windows host failed (dial refused, timeout, reset, TLS on/off mismatched with the listener, DNS) — the message names which and the fix (catalogue). An HTTP-level refusal by the host, or an untrusted or mismatched listener certificate, is IIS_AUTH, not this code. |
IIS_AUTH | auth / validation | The host answered 401/403, or an empty HTTP 500 on plain HTTP (AllowUnencrypted is false) — the message quotes the offered auth schemes, which identify the cause (walkthrough) — or the WinRM listener's TLS certificate is not trusted or was issued for another name (not retried). NTLM / Basic / Kerberos authentication rejected, an authorization denial before the bind ("Access is denied", 0x80070005, 401/403, "permission denied" — e.g. the account lacks store permissions), or a permanent Kerberos configuration error (qualified username, IP-addressed target, missing realm). A filesystem "Access to the path … is denied" is IIS_IMPORT_FAILED, not this code. |
IIS_PFX_FAILED | validation / io_transient | Building the PKCS#12 bundle locally failed (validation — cert/key problem), or uploading it over WinRM failed for a transport reason (io_transient — retried). A refused credential during the upload is IIS_AUTH, not this code. |
IIS_IMPORT_FAILED | io_permanent | A script failure before the bind step (module import, password file, opening or importing the PFX, binding lookup — including a filesystem access-to-path denial on the staged file), a missing thumbprint in the result, or a script result that could not be parsed. |
IIS_BINDING_NOT_FOUND | validation | No HTTPS binding matched the placement's filters on the selected site(s) — one site, a site list or every non-excluded site — or, with expect_single, none was left after the host-header filter. The message names the request, the binding a previous deploy updated that was changed or removed in IIS, and the site's HTTPS bindings now. |
IIS_BINDING_AMBIGUOUS | validation | More than one binding matched in single mode, or in a filtered selection with expect_single — refusing to guess; the message lists them. |
IIS_NAME_MISMATCH | validation | Verify Binding Names (opt-in) refused the binding: the certificate's names do not cover its host header, or — for a binding without a host header — share no name with the certificate currently on it. single mode fails the target; the multi-binding modes use this code only when every candidate (CCS bindings aside) was refused, otherwise they skip the binding and report partial. Also raised before any remote call when the certificate carries no DNS name. Nothing is imported or bound, so it does not trigger auto-rollback. Not retried. |
IIS_BINDING_FAILED | io_permanent | single mode: assigning the certificate to the binding threw (the exception message follows the code). Multi-binding modes: no bindings could be updated (including the all-matched-bindings-are-CCS case). |
IIS_APPPOOL_DETECT_FAILED | validation | Detecting the application pools to recycle failed. |
IIS_APPPOOL_RECYCLE_FAILED | io_permanent | Recycling one or more application pools failed (run is partial). |
IIS_VALIDATE_STORE | validation | Post-deploy check of the certificate store failed. |
IIS_VALIDATE_BINDING | validation / io_transient | Post-deploy binding check found a mismatch (validation), or the check itself failed at the transport level (io_transient — retried; the bind already succeeded). |
IIS_MODULE_MISSING | validation | The WebAdministration PowerShell module is missing (Install-WindowsFeature Web-Scripting-Tools). |
IIS_RESULT_UNREADABLE (warning, not an error code) | — | A remote script ran and exited 0 but its result could not be read, so the outcome of that step — an app-pool recycle, an old-certificate cleanup — is unknown. The deployment itself succeeded; re-run it, or check the host. |
IIS_HOST_PREREQ | validation | The host cannot run the deploy as it is. At deploy time: PowerShell in Constrained Language mode, a transfer larger than MaxEnvelopeSizekb allows (Windows answers HTTP 413), or a staging file in C:\Windows\Temp that cannot be created restricted to the connecting account (a leftover .captmp it cannot remove, or icacls failed — the file is removed). The health-check preflight also reports a non-administrator or UAC-filtered account, an unwritable C:\Windows\Temp, a PowerShell older than 3 and an inaccessible certificate store under this code — during a deploy those surface as IIS_AUTH or an import failure. The message names the fix (catalogue). |
IIS_SITE_NOT_FOUND | validation | A configured IIS site does not exist (for a site list the message names every missing site: site(s) not found: …), or the resolved placement names no site at all (a site-less legacy per-certificate override inherited onto a target that covers every site or a site list). Reported by Execute, Dry Run and the placement preview; nothing is imported. The health check reports a missing site as a warning (the target stays healthy). A missing site in excluded_sites is not this error — it is skipped with the warning IIS_EXCLUDED_SITE_NOT_FOUND. |
IIS_SITE_RENAMED | validation | A site the placement names (or an excluded site, renamed to a site the placement would bind) was renamed in IIS: the site a previous deploy recorded under that name — same site id, every recorded binding still on it, and a recorded binding with a host header or the same root folder — now has another name, which the message gives. Also makes the health check unhealthy. Nothing is imported; update the target or the override to the new name (details). |
Dual-class rows: the same code covers two failure natures; the module sets the class explicitly at the failure site, so fan-out retries only the transient one.
SMTP
The SMTP module classifies by SMTP reply code and sets the class directly on the result.
| Code | Class | Meaning |
|---|---|---|
SMTP_CONNECT | network | TCP dial / DNS failure before any SMTP conversation. |
SMTP_TLS | auth | TLS handshake or certificate-trust failure (STARTTLS/SMTPS). |
SMTP_AUTH | auth | Authentication rejected (reply 530/534/535/538). |
SMTP_TRANSIENT | io_transient | Temporary server condition (421/450/451/452/454, other 4xx, or errors without a reply code that are not a recognised negotiation failure). |
SMTP_PERMANENT | io_permanent | Permanent rejection (550/552/554, other 5xx) — operator action needed. |
SMTP_VALIDATION | validation | The request was malformed (500/501/503/553), or the target and the relay cannot agree deterministically: STARTTLS required but not advertised, AUTH or the chosen PLAIN/LOGIN mechanism not offered, password refused over plaintext. Not retried — change tls_mode/auth_method. |
SMTP_PAYLOAD | validation | Building the PEM / PKCS#12 attachment failed before sending. |
AWS ACM
The AWS ACM module classifies AWS SDK errors itself, so one code can map to more than one class depending on the underlying cause.
| Code | Class | Meaning |
|---|---|---|
AWS_CONNECT | network (dial/timeout) or io_transient (throttling) | AWS API unreachable, or the request was throttled (Throttling, RequestLimitExceeded, TooManyRequests, SlowDown). |
AWS_AUTH | auth | Credentials rejected (InvalidClientTokenId, SignatureDoesNotMatch, AccessDenied, UnauthorizedOperation, …). |
AWS_CREDENTIAL | auth | Resolving the module credential for the target failed. |
AWS_IMPORT | validation (bad input: missing region, ARN region ≠ target region, or an ARN naming a certificate ACM issued itself — refused before the import) or io_permanent (import rejected) | ImportCertificate failed or returned an unusable result, or the resolved ARN cannot be imported over. |
Cloudflare
The certificate is uploaded to a Cloudflare zone as a Custom Certificate (Business/Enterprise plan) over the Cloudflare API; renewals update it in place.
| Code | Class | Meaning |
|---|---|---|
CLOUDFLARE_CONNECT | network | Cloudflare API (api.cloudflare.com:443) unreachable, HTTP 5xx, or an HTTP 3xx redirect (never followed — a wrong or intercepting endpoint) — check egress/DNS/proxy and retry. |
CLOUDFLARE_AUTH | auth | API token rejected — verify the token is active and carries Zone → SSL and Certificates → Edit on the target zone. |
CLOUDFLARE_PLAN | validation | The zone's plan does not permit Custom Certificates — upgrade the zone to Business or Enterprise. Also raised when the zone's custom-certificate allocation is full (Cloudflare 1445 "Hit maximum cert allocation") — delete unused custom certificates or buy more slots. |
CLOUDFLARE_NOT_FOUND | validation | The referenced zone or custom certificate does not exist (or the zone is not visible to the token — Cloudflare's 9109 "Invalid zone identifier") — check the zone ID / certificate ID. |
CLOUDFLARE_CONFLICT | validation | The zone holds a custom certificate CertAutoPilot did not upload that covers the primary hostname, or — on a first upload — any other name on the certificate (Cloudflare serves the most recently uploaded certificate per name, so the upload would take it over; one is enough — there is no adoption by hostname) — take it over (certificate_id), upload alongside it (allow_overlap), or remove it in Cloudflare. Also raised for an HTTP 409 from Cloudflare. |
CLOUDFLARE_VALIDATION | validation | The request itself is invalid — bad bundle_method/type/geo_restrictions, or a cert/key the API rejected (any 4xx on the upload, including error 1002 there), or a zone holding more than 500 custom certificates in one state when a first upload or an id recovery needs the full listing. |
CLOUDFLARE_QUOTA | io_transient | API rate limit — HTTP 429, or error 1400 (the per-minute certificate-update limit, sent as HTTP 400) — retried; if persistent, slow the cadence. A full custom-certificate allocation is CLOUDFLARE_PLAN. |
CLOUDFLARE_HOST_OVERLAP (warning, not an error code) | — | A custom certificate CertAutoPilot did not upload also serves one of the certificate's names; the run went ahead (an update, a first upload with allow_overlap, or a first upload where the other certificate carries exactly a name this one covers only with a wildcard — never refused). The job log says which certificate the edge serves for each shared name. |
CLOUDFLARE_HOSTS_DROPPED (warning, not an error code) | — | The custom certificate that was updated in place served names the new certificate does not carry — typically the apex of a taken-over wildcard. It no longer serves those names (Cloudflare falls back to another certificate covering them, or its own edge certificate); add them to the certificate to keep them on it. |
cPanel
| Code | Class | Meaning & resolution |
|---|---|---|
CPANEL_CONNECT | network | The cPanel/WHM host is unreachable — check host, port, use_ssl and network egress; retried. |
CPANEL_AUTH | auth | The API token was rejected — verify username + token and the mode (cpanel vs whm). |
CPANEL_NOT_FOUND | validation | The server answered HTTP 404 — usually a wrong port/mode pairing (UAPI path on the WHM port or the reverse) or a proxy in front of the API. |
CPANEL_CONFLICT | validation | The server answered HTTP 409 — read the error text in the target detail and correct the server state. |
CPANEL_VALIDATION | validation | The target spec is invalid — check host, mode and the domain list. |
CPANEL_INSTALL | io_permanent | The API reported failure (status ≠ 1 in cpanel mode; metadata.result ≠ 1 or data.status ≠ 1 in whm mode) — check the certificate/key/chain and that the domain is on the account. |
CPANEL_QUOTA | io_transient | Rate limit exceeded — retried; if persistent, slow the cadence. |
Azure Key Vault
The certificate is imported into an Azure Key Vault as a new version of a KV certificate (AES-256-encrypted PKCS#12 under a random per-import password) over the Key Vault REST API, authenticated via an AAD service principal.
| Code | Class | Meaning |
|---|---|---|
AZUREKV_CONNECT | network | Vault or AAD endpoint unreachable (also raised on an unexpected redirect) — check egress and the vault_url. |
AZUREKV_AUTH | auth | Token rejected, missing permission, blocked by the vault firewall, or the owning Azure subscription is disabled. The message names the case — fix tenant/client/secret, grant Key Vault Certificates Officer (or access-policy Get+Import+List), allowlist CertAutoPilot's egress IP in the Key Vault firewall, or re-enable the subscription (no role change helps a disabled one). |
AZUREKV_NOT_FOUND | validation | Vault or certificate not found — verify the vault URL and certificate name. Also raised when the vault host does not resolve in DNS (a mistyped or deleted vault), which is permanent and therefore not retried. |
AZUREKV_CONFLICT | validation | A soft-deleted certificate holds the name (az keyvault certificate recover, or purge — blocked by purge protection — or pick a different cert_name); also covers a pending certificate operation. |
AZUREKV_VALIDATION | validation | Bad vault_url/cert_name, or Azure rejected the PFX. |
AZUREKV_QUOTA | io_transient | Key Vault throttling (429). Transient class: retried per target in a fan-out batch and by the sweep re-arm; the module itself performs no in-run backoff. |
MerlinCDN
Certificates on MerlinCDN are immutable — each deploy uploads a new cert, then binds it by re-pointing the distribution's canonical name. The codes below distinguish an upload failure (nothing bound yet) from a bind failure (a new cert was uploaded; the module best-effort deletes that orphan).
| Code | Class | Meaning |
|---|---|---|
MERLINCDN_CONNECT | network | MerlinCDN API unreachable. |
MERLINCDN_AUTH | auth | Personal access token or org/workspace headers rejected. |
MERLINCDN_NOT_FOUND | validation | The referenced distribution or canonical name does not exist. |
MERLINCDN_CONFLICT | io_permanent | Name/state conflict on the CDN. |
MERLINCDN_QUOTA | io_transient | API quota / rate limit hit (HTTP 429) — during the upload, or during binding when every failed row was rate-limited. Retried. |
MERLINCDN_UPLOAD | io_permanent | Uploading the new certificate failed (nothing was bound), or the upload landed but its id could not be recovered (a warning names the possible stray upload). |
MERLINCDN_BIND | io_permanent | Re-binding a canonical name was rejected — a 4xx other than 429, a missing CNAME/distribution, or an incomplete row. Not retried. If every row failed and the upload was fresh in this run, that upload is best-effort removed; if only some rows failed, the successful ones stay bound and a re-run heals the rest. A bind phase whose failures were all transient (5xx/network → MERLINCDN_CONNECT, 429 → MERLINCDN_QUOTA) is retried instead. |
Exchange
On-premises Microsoft Exchange, deployed over WinRM + the Exchange Management Shell. Every renewal imports a new thumbprint (import → enable → optional remove-old).
| Code | Class | Meaning |
|---|---|---|
EXCHANGE_CONNECT | network | WinRM unreachable — host/port/network — or a Kerberos domain controller that could not be reached (retried). A server's deploy script that loses its connection is re-run once within the same run (never once it restarts services itself). A domain controller that answered with a refusal is EXCHANGE_AUTH. |
EXCHANGE_AUTH | auth | The Exchange server refused the WinRM request (401/403, or an empty HTTP 500 on plain HTTP — AllowUnencrypted is false; the message quotes the offered auth schemes), or a TLS-verify / Kerberos misconfiguration (terminal). Exchange targets are Kerberos-only. |
EXCHANGE_EMS_UNAVAILABLE | io_transient | The Exchange /PowerShell RBAC session did not open (reason included) — not a real Exchange server, remote PowerShell disabled for the account, the account has no Exchange RBAC role, the per-user concurrent-session limit is exhausted, or the endpoint is down while IIS restarts. Retried: the last two clear on their own; a permanent cause fails again. A role that opens the session but lacks the certificate cmdlets surfaces as EXCHANGE_IMPORT / EXCHANGE_ENABLE instead. |
EXCHANGE_CREDENTIAL_MISSING | validation | No or malformed exchange_winrm credential (missing username/password). |
EXCHANGE_UNSUPPORTED_KEY_TYPE | validation | The certificate does not use an RSA key, or its RSA key is smaller than 2048 bits. Exchange accepts RSA 2048 and above only; re-issue the certificate with such a key. |
EXCHANGE_IMPORT | io_permanent | Import-ExchangeCertificate failed — or the certificate never reached the server because the host refused the upload for its size (MaxEnvelopeSizekb too low; the message says so and gives the command that restores the default). |
EXCHANGE_ENABLE | io_permanent | Enable-ExchangeCertificate failed. |
EXCHANGE_REMOVE | io_permanent | Reserved: defined but not currently emitted (a failed removal of the previous certificate is reported as a warning, not an error). |
EXCHANGE_NOT_FOUND | validation | Reserved: defined but not currently emitted. |
EXCHANGE_PFX_FAILED | validation | Building the PKCS#12 from the cert/key failed. |
PAN-OS
The certificate, key and chain are imported into the firewall as one named certificate object over the PAN-OS XML API, followed by a partial commit scoped to the API user.
| Code | Class | Meaning |
|---|---|---|
PANOS_CONNECT | network | TCP/TLS connection to the firewall management interface failed. Retried. |
PANOS_AUTH | auth | Authentication rejected (bad username/password or api_key — the firewall's 403 / code 16 "unauthorized" answer). Not retried. Note: username is required even with api_key. |
PANOS_IMPORT | io_transient | The certificate import was rejected or failed on the device. Retried — a same-name re-import is idempotent. Note: PAN-OS refuses a self-signed non-CA certificate, and the PKCS#12 passphrase is device-limited to 31 characters (the module already complies). |
PANOS_COMMIT | io_transient | The partial commit failed or its job timed out. Retried; a timed-out commit may still complete in the background, and the next run finishes with a commit-only pass instead of re-importing. |
PANOS_NOT_FOUND | validation | A referenced object does not exist on the firewall. Not retried. |
PANOS_VALIDATION | validation | The device rejected the request as invalid (bad name, vsys set while multi-vsys is off, malformed input). Not retried. |
PANOS_API | io_transient | An unclassified device-internal error, or a "Session timed out" answer (code 22) from a busy management plane — the same key keeps working, and a commit being polled when it happens is still waited for. Retried. |
A firewall that presents a certificate other than the target's pinned tls_fingerprint_sha256 is the cross-module TLS_TRUST_REFUSED (nothing sent, not retried), not PANOS_CONNECT.
FortiWeb
The certificate is updated in place on FortiWeb 8.0.7+, or imported as a new object per version and moved into SNI members and multi-certificate entries on earlier firmware. A server policy is never written unless the target's update_policies is on (FORTIWEB_POLICY_UPDATED).
| Code | Class | Meaning |
|---|---|---|
FORTIWEB_CONNECT | network | The request could not be exchanged with the management interface (TCP/TLS, reset, timeout) — also when every failed binding failed this way (the binding writes are safe to repeat). Retried. |
FORTIWEB_AUTH | auth | HTTP 401 — wrong username/password, CertAutoPilot's address not in the administrator's trusted hosts, the appliance refusing logins from CertAutoPilot's address after repeated failures (admin-lockout-threshold, counted per source address over every administrator), or the target's ADOM does not exist or is not assigned to the account (the appliance answers them all alike). Every target authenticates once without an ADOM first, so the message says whether the password or the ADOM was refused. Never retried; in a run, after a refused password no other target on the appliance with that username is attempted, after a refused ADOM only the targets naming that ADOM, and after one refused login fewer than the threshold nothing more is sent to the appliance (the message says "not attempted"). |
FORTIWEB_PERMISSION | auth | HTTP 421 — the access profile lacks the named group (sysgrp read-write, traroutegrp read). Not retried. |
FORTIWEB_IMPORT | io_transient | Import or in-place update of the certificate, or import of a CA certificate, failed. Retried — a re-run finds what an earlier attempt already did. |
FORTIWEB_API | io_transient | Unexpected device error. Retried. |
FORTIWEB_NOT_FOUND | validation | A referenced object does not exist (device -3). Not retried. |
FORTIWEB_VALIDATION | validation | The appliance refused the input — a reference to a name that does not exist (-651), a certificate in the wrong key slot (-39), material it cannot read or a key mismatch (-20019), an invalid argument (-20007), name rules — the certificate and key do not form a pair, an in-place update would change a multi-certificate slot's key type, or the target names an ADOM on an appliance without ADOMs. Not retried. |
FORTIWEB_PLACEMENT | validation | Some bindings were refused by the appliance (listed; the others were written), or none of the configured placements exist and nothing uses the certificate. Not retried. |
FORTIWEB_CHAIN_GROUP_MISSING | validation | A policy that serves the certificate has no intermediate group; serving the chain would need a policy change (it reloads the policy), so nothing was changed. Assign a group to the named policies once. Not retried. |
FORTIWEB_CONFLICT | io_permanent | The certificate name belongs to another CertAutoPilot certificate, or (before 8.0.7) a versioned name holds other content and is not this certificate's (its own object with another version gets the -<16 hex> name instead). |
An appliance that presents a certificate other than the target's pinned tls_fingerprint_sha256 is the cross-module TLS_TRUST_REFUSED (nothing sent, not retried), not FORTIWEB_CONNECT.
FortiGate
The local certificate is created once and updated in place afterwards; placements are added where missing on every deploy and never removed. A refused token is never retried, nothing more is sent to that firewall in the run, and the same token is not sent to that firewall again for about 70 seconds by the same CertAutoPilot process — refused tokens count toward the FortiGate admin lockout per source address (details).
| Code | Class | Meaning |
|---|---|---|
FORTIGATE_CONNECT | network | Connection, DNS or TLS failure, or an empty / non-JSON answer from the firewall. Retried. |
FORTIGATE_AUTH | auth | HTTP 401 — the token was refused (wrong, expired, revoked, or CertAutoPilot's address is outside the administrator's trusted hosts), the credential is malformed, or an earlier target of the run had its token refused ("not attempted"). Never retried. |
FORTIGATE_PERMISSION | auth | HTTP 403 — the admin profile lacks a right (vpngrp read-write; fwgrp address / others read-write for VIP / SSL inspection profile placements). Not retried. |
FORTIGATE_API | io_transient | Any other device error. Retried — every write is safe to repeat. |
FORTIGATE_VALIDATION | validation | A private key that does not belong to the certificate (checked before anything is sent), name longer than 35 characters, a factory certificate, the certificate the management interface serves (admin-server-cert), a certificate and key that do not match, the same certificate already carried by another object (named — set the certificate name to it to take it over, or delete it), a previous object of this certificate that still carries the version being deployed (a rollback after a rename — the message says what keeps it), an unknown VDOM (HTTP 424), a VDOM named while VDOMs are disabled, or a device answer that cannot change on a retry (a named object missing — HTTP 404 / -3, a full table -4, an object in use -23, a refused value -37, a CA certificate refused as invalid -146 / -324 — the evaluation licence refuses RSA CAs above 512 bits). Usually nothing is written; when it comes after a write (a CA import), the run reports the change. Not retried. |
FORTIGATE_PLACEMENT | validation | A named VIP or SSL inspection profile does not exist, or the profile is not in replace mode. Nothing was written. Not retried. |
FORTIGATE_CONFLICT | io_permanent | The certificate name holds another CertAutoPilot certificate's local certificate. Nothing was written. Not retried. |
A firewall that presents a certificate other than the target's pinned tls_fingerprint_sha256 is the cross-module TLS_TRUST_REFUSED (nothing sent, not retried), not FORTIGATE_CONNECT.
A10 Thunder
Certificate, key and chain are one file, replaced in place on every renewal; template and SNI entries are written where they do not name it yet and never removed. A refused login is never retried, and the same password is not sent to that device again for about 70 seconds. See A10 Thunder.
| Code | Class | Meaning |
|---|---|---|
A10_CONNECT | network | Connection, DNS or TLS failure, or no aXAPI answer from the device. Retried. |
A10_AUTH | auth | The username/password was refused (wrong password, unknown or locked administrator — the device gives one answer for all three), the credential is malformed, or the password is held after a refusal ("not attempted"). Not retried. |
A10_PERMISSION | auth | The administrator lacks the right: read-only, or a partition administrator outside its partition. Not retried. |
A10_API | io_transient | Any other device error, busy, or the device's session limit. Retried. |
A10_UPLOAD | io_transient | Uploading the file failed; uploading the same content again is safe. Retried. |
A10_TEMPLATE | io_transient | Writing a template entry or a template's SNI list failed. Retried. |
A10_VALIDATION | validation | Refused before anything is written: a certificate that is not RSA or ECDSA, a key that does not belong to it, a name or partition the device would not take, a template using server-name auto-map, an alternate / regex / passphrase SNI entry for a configured name, or a key-type change while a template naming the file also holds another file of the new type. Not retried. |
A10_CONFLICT | io_permanent | The certificate name is a file CertAutoPilot did not write (A10 files carry no owner), or a file / template entry another CertAutoPilot certificate on the target owns. Nothing was written. Not retried. |
A device that presents a certificate other than the target's pinned tls_fingerprint_sha256 is the cross-module TLS_TRUST_REFUSED.
Elchi
CertAutoPilot creates its own TLS secret in every version Elchi serves and updates only those. See Elchi.
| Code | Class | Meaning |
|---|---|---|
ELCHI_CONNECT | network | The controller could not be reached, or it answered a redirect. Retried. |
ELCHI_AUTH | auth | Login refused — wrong username or password, a one-time code required, a deactivated account. Not retried. |
ELCHI_PERMISSION | auth | The account cannot reach the project (not in a group of it) or cannot write (Viewer). Not retried. |
ELCHI_VALIDATION | validation | Elchi refused the certificate (its findings are listed — expired, broken chain) or the request, or there is no version to write to (the account sees no listener or cluster in the project — add it to the project's group). Not retried. |
ELCHI_CONFLICT | validation | A secret of this name exists and was not created by CertAutoPilot for this certificate — the target stopped before anything was written, in every version; set another secret name. Not retried. |
ELCHI_UNAVAILABLE | io_transient | Elchi could not reach its control plane for that version, or serves none of the project's versions. Retried. |
ELCHI_PUBLISH_FAILED | io_transient | The secret was saved but pushing it to a listener failed; the listeners are named. Retried — the next run writes and publishes again. |
ELCHI_PUBLISH_TIMEOUT | io_transient | The secret was saved but was still being published after 60 seconds. Retried — the next run writes and publishes again. |
ELCHI_API | io_transient | A server error, an answer CertAutoPilot could not read (is base_url the controller API address?), or the session was replaced again right after logging in (someone else logs in with the same account). Retried. |
Certificate issuers
Not distribution codes: these come back from the certificate and CA-connection APIs or land in a certificate's Last error. A refusal on certificate create is HTTP 400 with the code at the start of error; ISSUER_UNSUPPORTED_OPERATION is 422; the status of the others is given in the row. They have no error class and are never retried by fan-out.
| Code | Where | Meaning |
|---|---|---|
ISSUER_UNSUPPORTED_OPERATION | revoke / reissue / rekey API | The issuing CA cannot perform the operation from CertAutoPilot — Microsoft AD CS or an imported certificate (revoke), or a CA connector whose vendor cannot revoke (GlobalSign Atlas can; an active connector certificate can be re-issued only when the vendor reissues its order for free), or rekey on a certificate whose issuer offers none (only GlobalSign Atlas does). The body carries issuer_type, vendor and operation. |
ISSUER_REQUIRES_ZONE | create API | A CA-connector certificate has no zone_id; the vendor's validation record must be written into a zone. Not raised for a vendor that validates domains on its account (GlobalSign Atlas) — there the zone is optional. |
ISSUER_REQUIRES_DNS_CREDENTIAL | create API | The zone has no DNS credential, so the validation record cannot be written. |
CA_CONNECTOR_SAN_LIMIT | create API | More names than the connector supports per certificate (max_sans; GlobalSign Atlas has no connector limit — its account policy is checked before ordering, POLICY_MISMATCH). |
CA_PRODUCT_INVALID | create API · Last error | The product code is not on the connection, or the connection has no products; at renewal time, the certificate's product was removed from the connection — no substitute is ordered. |
CA_VENDOR_UNKNOWN | create / connection API | The connection names a vendor this build does not ship. |
CA_VENDOR_NOT_OFFERED | connection API | A new connection for a vendor that is no longer offered for new connections. Existing connections of that vendor keep working. |
CA_CONNECTION_NOT_ACTIVE | create API | The connection is flagged error (a failed test or an authentication failure during an order). Test connection (or a successful order) clears it. |
CA_CONNECTION_NO_CLIENT_CERT | connection API | Generate CSR on a connection whose vendor uses no TLS client certificate. |
CA_CONNECTION_CREDENTIAL_REQUIRED | connection API | The base URL changes without every stored secret entered again in the same request (the message names the missing fields) — a stored secret never follows a connection to a new address. |
CA_CONNECTION_ORDERS_PENDING | connection API | The base URL or auth scheme changes while orders placed at the current address are still pending. |
CA_VALIDATION_METHOD_UNSUPPORTED | create API | ca_domain_validation names a method the vendor does not offer (or the vendor validates no domain on its account). |
CA_VALIDATION_NOT_WAITING | e-mail / stop validation API (409) | No domain validation of this certificate is waiting — it finished, was stopped, or an order was already placed. |
CA_VALIDATION_BUSY | stop validation API (409) | A step of the certificate's order is running at that moment. Try again in a moment. |
CA_VALIDATION_EMAIL_NOT_ALLOWED | e-mail API (422) | The address is not one the vendor accepts approval e-mail for. |
CA_ACCOUNT_UNSUPPORTED | CA-connection account API (422) | The connection's vendor cannot describe its account. |
CA_DOMAIN_IN_USE | CA-connection account API (409) | A domain validation is still removed while a certificate waits for it. Stop that certificate's validation first. |
DOMAIN_VALIDATION_STOPPED | Last error | An operator stopped the domain validation. Nothing was ordered; Re-issue (first issuance) or Renew (renewal) starts again. |
ORDER_REQUEST_DISCARDED | Last error | A project admin discarded an order request the vendor never answered. Re-issue sends a new request; if the vendor did create a certificate for the discarded one, the new request uses quota for a second certificate. |
CA_ORDER_BUSY | discard order request API (409) | A step of the certificate's order is running, or the request is being asked again right now. Try again in a moment. |
CA_ORDER_NOT_UNCONFIRMED | discard order request API (409) | The order request was answered, has ended, or is not an order request — there is nothing to discard. Reload the certificate. |
CA_RENEWAL_THRESHOLD_TOO_SHORT | create / renewal settings API | The renewal window is not longer than the CA Order Deadline (default 72 h ⇒ at least 4 days). |
CA_RENEWAL_THRESHOLD_TOO_LONG | create / renewal settings API | The renewal window is not shorter than the longest validity the connection's account policy grants (read from the vendor; not checked when the policy cannot be read). |
CA_ORDER_DEADLINE_EXCEEDED | Last error | The vendor did not validate the domains or did not issue before the CA Order Deadline; the order (or the validation) was closed and its record removed. Reissue to try again. |
CA_ORDER_NOT_STORED | Last error | The vendor issued the order but CertAutoPilot could not store the certificate before the deadline. Do not order again: the job log names the order. |
CA connector codes
A connector failure writes <CODE>: <explanation> into Last error and into the code variable of the cert.issuance_failed / cert.renewal_failed notification. These codes are CertAutoPilot's own and apply to every vendor:
| Code | Meaning |
|---|---|
VENDOR_UNREACHABLE | The vendor API could not be reached (DNS, connection refused, TLS handshake, outbound policy) — nothing was sent. Retried by the job, the renewal ladder or the poll. |
VENDOR_NO_RESPONSE | The order request was sent and not answered usably (no answer, an HTTP 5xx). An order may exist. A vendor that can be asked again safely (GlobalSign Atlas) is asked again — it first looks for a certificate it already issued for the same key, so nothing is bought twice; otherwise CertAutoPilot never retries by itself — check the vendor panel. |
VENDOR_REDIRECT | The base URL answered with a redirect, which is never followed — set the final address. |
DCV_WRITE_FAILED | The DNS provider refused the validation record permanently. |
DOMAIN_NOT_VALIDATED | The vendor validates domains on its account, a domain is not validated there, and the certificate has no zone to publish the record in. Nothing was ordered. |
DOMAIN_VALIDATION_FAILED | The vendor did not accept the domain validation. Nothing was ordered. |
VENDOR_CERT_MISMATCH | The vendor's certificate does not name the ordered domains — refused, never stored. |
VENDOR_KEY_MISMATCH | The vendor's certificate does not carry the certificate's key — refused, never stored. |
VENDOR_CERT_INVALID | The vendor's certificate is expired, not yet valid, or (for a renewal) does not extend the current one — refused, never stored. |
CONNECTION_UNAVAILABLE | The connection could not be loaded while an order waited (a database or key-store blip); the poll tries again. |
Any vendor code CertAutoPilot does not recognise is treated as terminal for an order already sent, so it never places a second paid order by itself.
GlobalSign Atlas
GlobalSign Atlas answers with an HTTP status and a description; CertAutoPilot turns them into these codes. The full table with what to do is on the GlobalSign Atlas page.
| Code | Meaning |
|---|---|
AUTH_FAILED | Atlas refused the API key / secret, or the mTLS certificate is not linked to this API credential (HTTP 401/403). The connection turns error. |
MTLS_CERT_MISSING | The connection has no mTLS client certificate yet. |
MTLS_CERT_INVALID | The mTLS client certificate has expired, is not valid yet, or does not match its key. |
QUOTA_REACHED | The account's certificate quota is used up (HTTP 402). No certificate was ordered. |
POLICY_MISMATCH | The request does not fit the account's validation policy — key type or size, too many or too few names, a name outside the allowed suffixes, a required subject field, a required public key signature (HTTP 422, or CertAutoPilot's own check before sending). No certificate was ordered. |
REQUEST_INVALID | Atlas refused the request as malformed (HTTP 400). |
RATE_LIMITED | Too many requests (HTTP 429). Retried. |
SERVICE_BUSY | Atlas answered HTTP 5xx to a call that buys nothing. Retried. |
CERTIFICATE_NOT_FOUND | Atlas does not know the certificate (HTTP 404). |
CERTIFICATE_REVOKED | The certificate was revoked at Atlas before CertAutoPilot stored it. |
CERTIFICATE_FAILED | Atlas reports the certificate in a state other than issued or revoked. |
UNEXPECTED_ANSWER | An answer CertAutoPilot does not understand. |
CLAIM_CHECK_RUNNING | A domain validation was to be removed while Atlas was checking it (HTTP 202). Try again in a moment. |
Warning codes
Warnings do not fail a target and are never retried. They appear in the warnings list of the per-target result, next to a success (or partial) status, so a condition the operator should know about stays visible on an otherwise green run.
| Code | Module | Meaning |
|---|---|---|
OVERRIDE_GROUP_CONFLICT | Any | The target is in two or more of the module's target groups whose override rows all set a value, so no group value was applied: it used its own row, the Default row, its own settings or the generated name. Saving such rows is refused (422 with the same code); this arises when a target joins a second group afterwards. The dry run lists it before the run. |
K8S_NOTHING_MATCHED | Kubernetes | Find by certificate names: no TLS Secret in the namespaces searched holds the certificate's names, so nothing was written. |
K8S_SECRETS_SKIPPED | Kubernetes | Find by certificate names: Secrets that share a name with the certificate were left alone (another certificate, owned by cert-manager / Helm, immutable). The job log names each and why. |
K8S_OWNED_UPDATED | Kubernetes | Find by certificate names: a Secret managed by cert-manager, Helm or an owner reference was updated because include owned is on. |
K8S_ROLLOUT_SKIPPED | Kubernetes | Find by certificate names: a chosen workload did not roll out — paused, OnDelete, it uses no updated Secret, or the credential cannot list a namespace's workloads. |
K8S_NAMESPACE_FORBIDDEN | Kubernetes | Find by certificate names: a namespace in scope could not be read with the credential's permissions and was skipped. |
IIS_INSECURE_TRANSPORT | IIS | tls_skip_verify is on, so the WinRM HTTPS certificate is not verified. Plain HTTP (5985) does not raise it; with Basic auth that case only writes a job-log warning. |
IIS_LARGE_PFX | IIS | The generated PFX is larger than 20 KB. A large transfer is more likely to hit the host's MaxEnvelopeSizekb limit. |
IIS_TEMP_CLEANUP_FAILED | IIS | single mode only: the remote script could not delete the uploaded .pfx/.pw files. CertAutoPilot still deletes them afterwards. |
IIS_CCS_BINDING_SKIPPED | IIS | A multi-binding run skipped one or more Central Certificate Store bindings; they cannot be updated through the machine store. |
IIS_NAME_MISMATCH_SKIPPED | IIS | Verify Binding Names skipped one or more bindings this certificate does not serve in a multi-binding run. The other bindings were updated; the target is partial (reason binding_name_mismatch, unless a missing or failed binding set its own reason first). |
IIS_BINDING_NAMES_DROPPED | IIS | With Verify Binding Names on, a binding without a host header was updated, but the certificate previously on it also served names the new one does not cover. The run still succeeds. |
IIS_EXCLUDED_SITE_NOT_FOUND | IIS | A site in excluded_sites of an every-site placement is not on the server: skipped, the run went on (the job log and Dry Run name it). An excluded site renamed to a site the run would bind fails with IIS_SITE_RENAMED instead. |
IIS_PARTIAL_POOL_RECYCLE | IIS | Some application pools recycled and others failed. The certificate is bound; the target is partial with IIS_APPPOOL_RECYCLE_FAILED. |
IIS_ALL_POOL_RECYCLE_FAILED | IIS | Every pool recycle failed, or the recycle script itself failed. The certificate is bound; the target is partial with IIS_APPPOOL_RECYCLE_FAILED. |
IIS_OLD_CERT_IN_USE | IIS | remove_old_cert_after_bind kept a previous certificate because an SSL binding still uses it. |
IIS_OLD_CERT_CLEANUP_FAILED | IIS | Removing a previous certificate failed, or the cleanup script failed. The result status is unchanged. |
IIS_RESULT_UNREADABLE | IIS | A remote step (an app-pool recycle, an old-certificate cleanup) exited 0 but its result could not be read, so its outcome is unknown. Re-run the deployment or check the host. |
IIS_IPV6_BINDING_SKIPPED | IIS | Dry Run found IPv6 bindings. They cannot be targeted and were left out. |
CLOUDFLARE_HOST_OVERLAP | Cloudflare | Another custom certificate, not uploaded by CertAutoPilot, also serves one of the certificate's names (details). |
CLOUDFLARE_HOSTS_DROPPED | Cloudflare | The custom certificate updated in place served names the new certificate does not carry, so it no longer serves them (details). |
TLS_UNVERIFIED_PASSWORD | PAN-OS, F5 BIG-IP, NetScaler, FortiWeb, FortiGate, A10 Thunder | The target's password credential (FortiGate: its API token) was sent to the appliance without verifying its certificate: ssl_verify is off and no tls_fingerprint_sha256 is pinned (over HTTPS for F5/NetScaler; a PAN-OS API-key credential is not warned). Raised by deploy, dry run and health check, naming the served certificate's SHA-256. Clear it by pinning that fingerprint, or by turning ssl_verify on (with ca_cert for a private CA) — see the module page's Trusting the appliance's certificate. |
F5_PROFILE_SKIPPED | F5 BIG-IP | At least one listed client-ssl profile does not exist on the device and was skipped. It repeats on every run until the profile is created or removed from profile_names. |
PANOS_COMMIT_PENDING | PAN-OS | auto_commit is off, so the certificate was imported into the candidate configuration only. The firewall keeps serving the old certificate until an administrator commits. |
FORTIWEB_PLACEMENT_SKIPPED | FortiWeb | A configured SNI group or member, or multi-certificate entry does not exist on the appliance — skipped, the others were bound. Repeats until it is created or removed from the target. |
FORTIWEB_POLICY_BINDING_SKIPPED | FortiWeb | Firmware before 8.0.7: a server policy still uses the previous version in its own certificate field. It is not changed (that reloads the policy and drops its connections); bind it through a multi-certificate entry, or upgrade to 8.0.7+ where the certificate is updated in place. |
FORTIWEB_OLD_CERT_KEPT | FortiWeb | A previous version stays on the appliance — something the module does not change still uses it, or it was not created by this certificate. |
FORTIWEB_NOT_BOUND | FortiWeb | The certificate is on the appliance but nothing uses it yet. Its intermediate group (cap-icg-<8 hex>, or the target's chain group) was created with the chain — bind them once: on 8.0.7+ select the certificate and that intermediate certificate group in the policy (or use an SNI member / multi-certificate entry); on 8.0.5–8.0.6 bind it through a multi-certificate entry or an SNI member (a policy's own certificate field would stay on this version). |
FORTIWEB_ADOPTED | FortiWeb | An existing certificate of this name, not created by CertAutoPilot, was taken over (8.0.7+): updated in place, or — when it already carried this version — only tagged as owned by this certificate. |
FORTIWEB_PLACEMENT_UNUSED | FortiWeb | The certificate was bound to an SNI group no server policy has enabled, or a multi-certificate entry no policy or SNI member uses — no client receives it there. Pick a place a policy uses, or enable it on a policy. |
FORTIWEB_CHAIN_GROUP_SHARED | FortiWeb | An intermediate group serving the certificate also carries CA certificates outside its chain (added by another certificate or by hand). FortiWeb sends every member, so those clients receive them too. SNI members are moved to the certificate's own group automatically (unless that group is the target's explicit chain group); a policy's group cannot be changed without reloading the policy — give that policy a group of its own. |
FORTIWEB_TARGET_MOVED | FortiWeb | The target now points at another ADOM (or another appliance) than the one this certificate was last deployed to for it — the target was edited. The certificate is deployed there from scratch; nothing in the earlier ADOM or on the earlier appliance was changed or removed, so its objects keep serving the earlier version until removed by hand (the job log names the ADOM and object). |
FORTIWEB_POLICY_UPDATED | FortiWeb | The target's Also update server policies (update_policies) is on and a server policy was written — its own certificate field moved to the new version (8.0.5–8.0.6, at every renewal) and/or the certificate's intermediate group assigned to it (once). FortiWeb reloaded the policy and dropped its established connections. |
FORTIGATE_ADOPTED | FortiGate | The certificate name pointed at a local certificate CertAutoPilot had not created; it was updated in place and tagged as this certificate's object. |
FORTIGATE_NOT_BOUND | FortiGate | The certificate is on the firewall but no VIP, SSL inspection profile or SSL VPN uses it and the target names no placement — bind it once on the FortiGate; renewals update the same object, so the binding stays. |
FORTIGATE_TARGET_MOVED | FortiGate | This certificate was last deployed for the target to another firewall (serial) or VDOM — the target was edited. Nothing there was changed. |
FORTIGATE_OLD_CERT_KEPT | FortiGate | A previous local certificate of this certificate (the certificate name changed) stays on the FortiGate: something still uses it, so it keeps serving the earlier certificate (warned on every run, whatever delete_old_cert says), or a rename waits for the next renewal because the previous object already carries this certificate. |
FORTIGATE_CONNECTIONS_RESET | FortiGate | An intermediate CA certificate the firewall did not hold was imported. About 4 seconds later FortiGate closed the TLS connections it terminates (SSL offload on any VIP) that were open at that moment; their clients reconnect. New connections and other firewall traffic are not affected. A CA the firewall already holds is reused without this; renewals under the same intermediate import nothing. |
A10_NOT_BOUND | A10 Thunder | The file is on the device but no virtual port uses a template that names it — name a template a virtual port uses, or bind it once on the device; renewals replace the same file, so the binding stays. |
A10_ADOPTED | A10 Thunder | A template entry (or SNI entry) named a file CertAutoPilot had not written; it now names this certificate. The other file stays on the device. |
A10_SNI_NOT_COVERED | A10 Thunder | An SNI entry names a server name the certificate does not cover; that entry was left as it is. |
A10_OLD_CERT_KEPT | A10 Thunder | A previous file of this certificate (the certificate name changed) stays: a template still names it, or delete_old_cert is off. |
A10_TARGET_MOVED | A10 Thunder | This certificate was last deployed for the target to another device or partition — the target was edited. Nothing there was changed. |
A10_CONFIG_NOT_SAVED | A10 Thunder | write memory failed: the change runs but is lost on a reboot until the configuration is saved. The next run saves again, even when it writes nothing else. |
A10_HA_UNKNOWN | A10 Thunder | aVCS or VRRP-A is configured; CertAutoPilot writes to the unit that answers and does not synchronise the peer. |
ELCHI_NOT_BOUND | Elchi | The secret was saved, but no listener references it in that version yet — bind it in Elchi (a Downstream TLS context that names it). |
ELCHI_RECREATED | Elchi | The secret CertAutoPilot created had been deleted in Elchi and was created again under the same name. |
ELCHI_TARGET_MOVED | Elchi | The target now points at another controller or project; the secrets there were left untouched and new ones were created here. |