* Add UniFi OS Server deploy hook
Uses UniFi OS Server's local REST API (login, list, upload, activate,
remove superseded) since it stores certificates in its own Postgres
database rather than flat config files, unlike the Cloud Key/UDM
hardware covered by the existing unifi deploy hook. Tested against
real instances on both macOS and Ubuntu 26.04 (self-hosted, remote).
* Address review: portable sed/grep, scoped HTTPS_INSECURE, fingerprint matching
- Replace GNU-only \n in sed replacement with a portable literal newline
(matches dnsapi/dns_cpanel_uapi.sh, dnsapi/dns_glesys.sh); pipe the
list response through _normalizeJson first for consistent formatting.
- Use grep -F for the domain-name match instead of an unescaped BRE --
a wildcard cert name (*.example.com) broke the regex.
- Drop \W (undocumented, GNU-only) from the cookie lookup in favor of
an anchored `^Set-Cookie: *NAME=` match.
- Scope HTTPS_INSECURE=1 inside the hook (matches deploy/proxmoxve.sh,
deploy/fritzbox.sh) instead of requiring the caller to export it for
the whole acme.sh run, which would also disable verification for the
connection to the ACME CA.
- On a duplicate-certificate response, match the existing entry by
fingerprint instead of taking the first name match -- with more than
one stale entry for a domain, the wrong one could get activated.
- Check the list endpoint's response code before proceeding.
- Save username/password with the "base64" flag (matches
deploy/synology_dsm.sh) since _save_conf wraps values in unescaped
single quotes.
* Rework certificate handling: unique names per upload, drop cleanup
Testing against a real UniFi OS Server showed the server enforces name
uniqueness independently of fingerprint uniqueness, and that activation is
exclusive server-wide regardless of name/domain. A unique name per upload
avoids the name-collision path entirely (previously only handled as a
retry-of-identical-content edge case), and removes the need for the
post-hoc cleanup loop, which risked deleting the wrong entry.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
* Shorten generated certificate name to Unix epoch seconds
Real-hardware testing showed the UniFi OS Server certificate list's name
column is fixed-width and doesn't wrap, so a full human-readable timestamp
overlaps the Expires column and makes both unreadable. Epoch seconds are
still short enough to fit while remaining unique.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
* Add scoped cleanup of old certificate entries, use _time helper
Per review: dropping cleanup entirely went further than the original bug
required, and left old entries (each holding a private key) accumulating
indefinitely. Since every upload now gets a name unique to its domain and
run, cleanup can safely target only entries whose name starts with that
domain -- entries this hook itself created -- excluding the one just
activated. Also swaps date +%s for the core _time helper, and rewrote the
design comments to make them clearer and match the current behavior
instead of the pre-redesign one.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
Ubuntu / Ubuntu (Smallstep Intermediate CA, Smallstep Intermediate CA, , 1, https://localhost:9000/acme/acme/directory, 1, , 172.17.0.1) (push) Has been cancelled
_getdeployconf assigns and exports the variable, it does not print the
value, so wrapping it in a command substitution ran it in a subshell and
always yielded an empty string. A MULTIDEPLOY_FILENAME saved by an
earlier run was therefore never restored on renewal and the hook
silently fell back to multideploy.yml. Call it the same way every other
deploy hook does.
Also treat a MULTIDEPLOY_FILENAME starting with '/' as an absolute path
instead of always resolving it under DOMAIN_PATH, so one deploy file can
live outside the certificate directory and be shared by all domains.
Names without a leading '/' keep resolving under DOMAIN_PATH as before.
_temp_admin_cleanup ran before _logout, so the logout request carried
the session id of an account synouser had already removed and DSM kept
the orphaned entry in Connected Users. Swap the order in both terminal
branches, and add the missing _logout to the two post-login error paths
(CRT list failure, certificate not found without SYNO_CREATE).
_logout overwrites the global $response, so the upload-failure branch
prints its error message before calling it.
Reported by @Bertl75 in #7174
Solaris /usr/bin/grep has no -A ("illegal option -- A"), so _getAKI
printed an error to stderr on every cron renewal and returned empty.
The empty AKI silently corrupts the RFC 9773 ARI certID, so ARI is
never available and renewal falls back to the fixed schedule.
Split the pipeline into a testable stdin filter _extractAKI and select
the value line with a portable sed range instead.
Same fix for the two hooks that still used grep -A: dns_world4you.sh
(also replaces the GNU-only "\s" in the same expression) and
deploy/keyhelp.sh (the -A 2 window could truncate the div range that
follows it, so it is just dropped).
https://github.com/acmesh-official/acme.sh/issues/7159
* feat: add Shelly Gen3+ deploy hook with RFC 7616 HTTP Digest auth
Adds deploy/shelly.sh for deploying Let's Encrypt HTTPS server certificates
to Shelly Gen3+ devices (Gen4 tested) via JSON-RPC over HTTP.
- RFC 7616 SHA-256 HTTP Digest authentication (Authorization header)
- Uploads fullchain.pem and private key via Shelly.PutHTTPServerCert / PutHTTPServerKey
- Auto-reboot support (SHELLY_REBOOT to disable)
- Auth auto-detection: no password = no auth, password = Digest
- Nonce counter (nc) increments per request per RFC 7616
- Tested against Shelly 2PM Gen4 (firmware 2.0.0)
Also adds deploy/test_shelly.sh for self-testing the hook logic without
a real device (mocked _post).
* fix: address review feedback on shelly deploy hook
- Fix _secure_debug calls to use two arguments (label + value)
- Remove bash-only $RANDOM cnonce fallback; openssl always available
- Parse $HTTP_HEADER directly instead of raw curl re-request
- Detect auth via HTTP 401 status line, not empty response body
- Route reboot through _shelly_rpc to rebuild auth header with correct nc
- Remove export HTTPS_INSECURE=1 (no-op for http://, leaks to other hooks)
- Clear _H1 before returning from shelly_deploy
- Prefix all helper variables with _shelly_ to avoid namespace collisions
- Delete deploy/test_shelly.sh (deploy/ files become hook names)
- Fix missing trailing newline
* fix: validate shelly JSON-RPC responses are valid JSON
Non-JSON responses like HTTP 429 'Too Many Requests' would pass
the empty-response and '"error"' checks and be reported as success.
Now reject any response that doesn't start with '{' and contain '"id"'.
* fix: add 1s delay between shelly cert/key clear and upload calls
The Shelly device has a race condition where uploading data immediately
after clearing the existing cert/key returns -103 'Missing required
argument data!'. A 1-second delay fixes this.
* fix: remove clear-before-upload in shelly deploy hook
Shelly auto-removes all three TLS files (cert, key, CA bundle) when any
single one is cleared. The old sequence clear-cert → upload-cert →
clear-key → upload-key resulted in the key clear wiping the newly
uploaded cert, leaving only the key at boot time. The mbedtls
pk_check_pair then silently skipped the HTTPS listener.
Fix: just upload directly (overwrite in place). No clearing needed.
* Fix ShellCheck SC2090 and shfmt in shelly deploy hook
SC2090: false positive on export _H1 (used quoted in _post)
shfmt: no space after "<" in _json_encode redirects
* moved two lines to cover the whole if block
---------
Co-authored-by: neil <github@neilpang.com>
Co-authored-by: cysimons <cysimons@cisco.com>
The success check only grepped "message" from the response body, but
PVE/PBS auth failures return HTTP 401 with an empty body, so wrong or
unauthorized API tokens were reported as "Certificate successfully
deployed". Also _retval captured the exit code of the message pipeline
instead of _post. Check the HTTP status line from $HTTP_HEADER and
capture _post's exit code directly.
fix https://github.com/acmesh-official/acme.sh/issues/7141
Ubuntu / Ubuntu (Smallstep Intermediate CA, Smallstep Intermediate CA, , 1, https://localhost:9000/acme/acme/directory, 1, , 172.17.0.1) (push) Waiting to run
dnsapi/deploy: remove POSIX character classes from sed/grep patterns
Solaris /usr/bin/sed and /usr/bin/grep parse [[:space:]] etc. as a
literal bracket set and silently mis-match. Replace with [ ]* for
JSON matching, a printf-tab bracket for user-input trimming, and
[0-9] for digits; also drop GNU-only sed -r/-E in rage4, selfhost
and selectel, and reuse _strip_blank_lines in byteplus_alb.
* Deploy certificate to FortiGate firewall using API
* Refactor FortiGate deployment functions
Prefix private functions and working variables and use a timestamped certificate name.
* Replace grep -o for POSIX compatibility
* add Baidu Cloud CDN deploy hook
Code generated by GitHub Copilot with Claude Sonnet 4.6. Tested with local environment by human.
* inline functions
Code generated by OpenAI Codex with GPT-5.5 Sol. Tested with local environment by human.
The hook is sourced by acme.sh, so the bash shebang never takes
effect: under dash, `[ x == y ]` fails with "unexpected operator",
the 403 branch never triggers and 2FA-OTP login is skipped.
Replace `==` with `=` and use the standard sh shebang.
Ubuntu / Ubuntu (Smallstep Intermediate CA, Smallstep Intermediate CA, , 1, https://localhost:9000/acme/acme/directory, 1, , 172.17.0.1) (push) Waiting to run
With --signcsr the private key never exists in the cert home, so every
renewal printed ".../domain.key: No such file or directory" from the
shell redirection. Skip the key read in that case; the install_ssl call
already ran with an empty key there and cPanel keeps the installed one.
https://github.com/acmesh-official/acme.sh/issues/6228
grep -o '[0-9]*' can match the empty string; GNU grep skips empty
matches but BSD greps handle them differently, breaking the 2FA
login flow on OpenBSD. Force a non-empty match at all three sites.
from https://github.com/acmesh-official/acme.sh/pull/6725
Mirrors _clearaccountconf_mutable: clears the SAVED_ prefixed key and
the legacy unprefixed key. Replaces the local copy in synology_dsm.sh
and the direct _cleardomainconf call in multideploy.sh.
Closes#4722. Thanks to @sg1888.
ssh_deploy() ignored the result of _ssh_deploy and always returned
success, so a failed transfer to one (or all) of the servers in
DEPLOY_SSH_SERVER was silently swallowed. Track the return code across
the loop and return non-zero if any server failed, letting the caller
handle notification.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Committing after a failed import leaves a mismatched cert/key pair on
the firewall (PAN-OS does not validate the pair at commit time), which
can lock the admin out of the https management interface.
The docker deploy hook copied the key file preserving the source mode
(root:root 0600), so a non-root container service (uid >= 1000) could not
read it. Add DEPLOY_DOCKER_CONTAINER_KEY_MODE and
DEPLOY_DOCKER_CONTAINER_KEY_OWNER, applied via chmod/chown inside the
container after the key is copied and before the reload command.
Closes#5333
dash's echo has no -e flag and sends a literal "-e " prefix to the
socket, so haproxy rejects the command and the hot update always fails
on Debian/Ubuntu (/bin/sh = dash). Also accept "Transaction updated",
which haproxy replies when an uncommitted transaction already exists.
fix https://github.com/acmesh-official/acme.sh/issues/6165
- Removed scope exclusion for "standard commit".
- If 'device-and-networks' is excluded (previous behaviour), a certificate for Panorama (always outside of a template) will not be committed (imported to the config but never applied to Panorama). Therefore, panos.sh was only working for certificates used in templates and applied to devices, but not for the Panorama certificate itself.
- According to the official documentation and the XML API Browser, there is no 'policy-and-objects' that can be excluded.
- Although it is not mandatory that the user account is solely dedicated to replace certificates and to perform no other type of operations, it is recommended. If such recommendation is applied, the only changes being committed would be in relation to certificates. Therefore, it should be safe not to exclude any scopes.
- Changed the order for "force commit" from '<commit><partial><force>' (unofficial) to '<commit><force><partial>' (official). Both work, but it is recommended to use what is part of the official documentation and/or XML API Browser.
- Removed unofficial 'policy-and-objects' from commented out code (see above).
- Replaced 'exclude' with 'excluded' from commented out code, as per the official documentation. Both work, but see above.
- Replaced 'acmekeytest' with $_panos_user in the commented out code.
Official documentation: https://docs.paloaltonetworks.com/ngfw/api/pan-os-xml-api-request-types-and-actions/commit
XML API Browser: https://<PANOS HOST>/api
