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
OmniOS native egrep -o infinite-loops emitting empty lines when the
pattern can match the empty string, so `_egrep_o "[^:]*$"` never lets
the pipeline finish and dns_yc hangs until the CI timeout. Require at
least one character instead. `+` is not usable because the sed fallback
in _egrep_o parses BRE.
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 $HTTP_HEADER and
capture _post's exit code directly.
fix https://github.com/acmesh-official/acme.sh/issues/7141
When ACME_PACKAGED is set (e.g. exported by a distro package wrapper):
- --install does not copy the script or the hooks into LE_WORKING_DIR;
the cron job and the shell alias point to the packaged script instead
- --upgrade, --install-online and the cron AUTO_UPGRADE path refuse and
point to the system package manager
- --uninstall does not remove the packaged files
https://github.com/acmesh-official/acme.sh/issues/7135
dnsapi/deploy: remove POSIX character classes from sed/grep patterns
Solaris /usr/bin/sed and /usr/bin/grep parse [[:space:]] etc. as a
literal bracket set and silently mis-match. Replace with [ ]* for
JSON matching, a printf-tab bracket for user-input trimming, and
[0-9] for digits; also drop GNU-only sed -r/-E in rage4, selfhost
and selectel, and reuse _strip_blank_lines in byteplus_alb.
A negative --days is anchored to the expiry, so it composes with a
relative --valid-to: "--valid-to +30d --days -7" renews 7 days before
the expiry instead of the hardcoded 1 day. A positive --days and any
--days with a fixed-date --valid-to are still rejected.
https://github.com/acmesh-official/acme.sh/issues/6570
_egrep_o falls back to a BRE sed expression on shells without egrep -o
(Solaris, DD-WRT busybox). A bare "\{" there is a BRE interval operator
and aborts ("sed: command garbled" / "Invalid content of \{\}"), so the
challenge-status-invalid path extracted an empty error object and the
CA's failure reason was lost. Replace the escaped braces with "[{]"/"[}]"
bracket expressions, unambiguous literal braces in both BRE and ERE, at
all four call sites (challenge type/error extraction and profiles).
When re-registering an already-bound EAB account (HTTP 400 "not
awaiting external account binding"), the else branch restored
ACCOUNT_URL from ca.conf but the following unconditional
`export ACCOUNT_URL="$_accUri"` clobbered it with an empty _accUri
(never set on that path), so later signed requests failed with
"A Key ID MUST be specified" / "account URL is empty". Assign the
restored value to _accUri so the shared export uses it.
https://github.com/acmesh-official/acme.sh/issues/3382
The sed in _get_zone passed the raw JSON through when the response
contained no [{"fqdn":", so an API error turned the zone into "{" and
the failure only surfaced later as POST /2/zones/{/records
"method_not_found". Log the response at debug2, error out on
non-success results, and parse fqdn position-independently.
https://github.com/acmesh-official/acme.sh/issues/6851
The "no crontab" whitelist used multiple -e patterns, but Solaris
/usr/bin/grep honors only a single -e, so a fresh install was
refused there. Use one plain grep per message pattern, which every
grep implementation supports (caught by le_test_installcronjob_no_wipe
on the Solaris CI).
The zone cannot be derived from the record name when the Knot server
is only authoritative for a delegated subdomain; let the user name it
explicitly, like NSUPDATE_ZONE.
fixes https://github.com/acmesh-official/acme.sh/issues/2881
The legacy plain _saveaccountconf copy in account.conf is sourced at
startup and silently overrides a newly exported DNSimple_OAUTH_TOKEN,
so rotated tokens never took effect.
fixes https://github.com/acmesh-official/acme.sh/issues/3422
Piping a failed 'crontab -l' straight back into 'crontab -' replaced
the whole crontab with just the acme.sh entry when the listing failed
while jobs existed (seen on cPanel/CloudLinux jailshell). Capture the
listing first and refuse to write unless the failure is the normal
"no crontab for user" case.
https://github.com/acmesh-official/acme.sh/issues/3079
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.
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
Debug logs occasionally contain private keys or tokens (issue 6267);
the code-side leak in the haproxy hook was fixed by #6268, this adds
the missing warning to the auto-comment that asks for logs.
The default schedule was a fixed CertCreateTime + RenewalDays - 1 day,
which passes notAfter entirely for short-lived certs (internal CAs
today, the CA/B SC-081 47-day maximum later) and leaves an expired cert
in place for weeks. Extract the arithmetic into _calc_next_renew_time
and cap it at one day before expiry (one hour for lifetimes of 24h or
less, mirroring --valid-to scheduling). CAs with ARI are unaffected --
the ARI window still overrides afterwards.
https://github.com/acmesh-official/acme.sh/issues/6305
The wiki has documented "21. Set notification for customscript" since
2022 but the implementation (#4193) was never merged, so following the
wiki failed with "Cannot find the hook file". Same interface as
documented: the script gets subject, content and status code as three
arguments. Unlike #4193, the target script is invoked directly instead
of through eval -- the subject/content contain domain names and CA
messages, and eval would allow command injection through them.
https://github.com/acmesh-official/acme.sh/issues/6377
The unfiltered GET /zones lists every zone on the server; with large
installations (100k zones) root-zone detection took minutes per domain.
Probe each walk-up candidate with ?zone=<name> instead (exact match per
the PowerDNS API docs); servers that ignore the parameter return the
full list, which the existing check still handles.
https://github.com/acmesh-official/acme.sh/issues/6382
Missing socat only matters for standalone mode; the text even says the
warning can be ignored. Printing it to stderr made every --upgrade in a
cron noisy for DNS-only users who redirect stdout.
https://github.com/acmesh-official/acme.sh/issues/6525
The dns/deploy hooks export _H1.._H5 in the main process, and the
notify hooks run in a subshell that inherits them. A hook that does
not overwrite every slot (ntfy without NTFY_TOKEN, slack, telegram,
etc.) sent the stale headers with its request, leaking another
service's Authorization credentials to the notify endpoint.
https://github.com/acmesh-official/acme.sh/issues/6801
With both domain.tld and sub.domain.tld zones on the account, the first
endswith hit could pick the parent zone while cPanel stores the record
in the most specific one, so the cleanup never found the record and
left an orphaned _acme-challenge TXT entry. Pick the longest matching
zone with an exact literal suffix match (_endswith treats the needle as
a regex, letting xdomain.tld wrongly match zone domain.tld).
https://github.com/acmesh-official/acme.sh/issues/6807
The Njalla API returns record ids as JSON strings now; the numeric-only
pattern matched nothing, so the removal never found the record id.
Match both quoted and bare ids.
from https://github.com/acmesh-official/acme.sh/pull/5121
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
A trailing dot in --domain-alias/--challenge-alias was passed through
to the dnsapi hook verbatim. Providers with exact-match record-name
lookups (e.g. Cloudflare's name= filter) then never find the record,
so rm never deletes it and relic TXT records accumulate on every issue.
Stripping in issue() also fixes certs with a dotted alias already
saved in domain.conf.
fix https://github.com/acmesh-official/acme.sh/issues/4636
LC_ALL in the environment overrides both LC_TIME and LANG, so LANG=C
alone still produced localized day/month names on non-English systems
and DNS Made Easy rejected the request date header. An LC_ALL=C
command prefix beats every locale variable (same pattern as
dns_oci.sh).
Fixes#4272. Closes#4271. Thanks to @Nickinthebox.
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.
Postfix with smtpd_forbid_bare_newline (default hardening since 3.9,
after SMTP smuggling) rejects the message with
"521 5.5.2 Error: bare <LF> received". RFC 5321 requires CRLF.
The python sender is unaffected (smtplib already emits CRLF).
fix https://github.com/acmesh-official/acme.sh/issues/7104
OpenBSD grep treats \| in a BRE as a literal | character, so
_freedns_domain_id never matched any row and every domain lookup
failed with "Domain not found". Switch to ERE with -E, keeping the
parens escaped so the (.*) suffix branch still requires literal
parentheses and does not widen the match (e.g. searching example.com
must not match example.company).
Reported-by: @katiekloss @boretom
Ref: https://github.com/acmesh-official/acme.sh/issues/2305
ACCOUNT_EMAIL / --email now accepts a comma- or space-separated list
and registers all of them as ACME contact entries. The ZeroSSL EAB
endpoint takes a single address, so the first one is used there.
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.
aws_ses_send calls `_use_container_role || _use_instance_role` when no
static AWS keys are set, but those functions were never defined -- only
_use_metadata was -- so role-based auth silently fell through to the
"no api key" error. Add both, using the current IMDSv2-capable versions
from dns_aws.sh, and set the IMDSv2 token header in _use_metadata so the
credential fetch works on IMDSv2-only instances.
Closes#4742
_get_root matched a registered domain anywhere as a substring of the
challenge domain, so with both "test.com.ru" and "subtest.com.ru" in the
account, issuing for subtest.com.ru wrongly picked test.com.ru as the
root (it is a substring of "sub-test.com.ru"). Anchor the match to a '.'
boundary so a shorter domain no longer matches a longer subdomain label.
Fixes the issue reported in #5036 (thanks @koledas)
Closes#5036
The domain list is paginated at 1000 entries per page and only the
first page was fetched, so accounts with more than 1000 domains never
found the root zone.
fix https://github.com/acmesh-official/acme.sh/issues/5051
An undecodable key (e.g. broken LibreSSL base64 -d -A) used to produce
the cryptic "Usage: _hmac hashalg secret [outputhex]" and an empty EAB
signature that the CA rejects with 403.
https://github.com/acmesh-official/acme.sh/issues/4082
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
Runs le_test_nginx from acmetest against Pebble: nginx listens on
Pebble's HTTP-01 validation port with an aaPanel/BT style
"location ^~ /" reverse proxy block, the regression case of #6125.
- replace deprecated set-output with GITHUB_OUTPUT
- untap aws/tap before brew install to silence tap trust warning
- inject safe.directory=* for cygwin git so the checkout post step
no longer fails with dubious ownership (exit 128)
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
"openssl req -noout -in" aborts when the default config file is missing;
reading a CSR needs no config, so pass -config /dev/null explicitly.
Stock NetBSD does not install /etc/openssl/openssl.cnf, so --signcsr
never worked there.
With a misconfigured $HOME / CERT_HOME the glob over "$CERT_HOME"/*.*
matches nothing, so renewAll silently does nothing and returns success --
--renew-all / --cron appears to work while renewing no certificates.
Check that CERT_HOME is a directory up front and return 1 with a clear
error instead.
Closes#4508
Core-Networks' API rejects Unicode domain names with "invalid domain";
it requires punycode. dns_cn_add / dns_cn_rm passed the raw challenge
domain straight through, so IDN certs failed at the TXT add step
(issue #4804). Run fulldomain through _idn() in both functions. For
ASCII/punycode input _idn() is a pass-through, so non-IDN domains are
unaffected.
Fixes#4804
The unquoted splitting let a "*" segment expand against files in the
current directory, so "*.*.*.*" could pass as a valid IPv4 address
(issue 4971). The old code also accepted "", "1.2.3", "1.2.3.4.5",
"1..2.3" and bare numbers. Split with IFS under set -f, require 4
octets, and validate each as a 1-3 digit number <= 255.
Based on https://github.com/acmesh-official/acme.sh/pull/4974
fix https://github.com/acmesh-official/acme.sh/issues/4971
_get_root matched the candidate zone with _contains (grep), which treats
the domain as a regex. For "-d g.<zone>" the candidate "g.<zone>" matched
"<string>...<zone>" because '.' matches the '>' after "string" and the 'g'
comes from the "<string>" tag, so "g.<zone>" was wrongly taken as the root
zone (sub=_acme-challenge instead of _acme-challenge.g). Anchor the match
to <string>$h</string> and escape dots so the zone is compared literally.
Fixes#5129
RFC 8555 sec 7.3.6 requires 401 (Unauthorized) when a request is
signed by a deactivated account, which ZeroSSL follows, while
Boulder (Let's Encrypt) historically returns 403. Check both codes
in _regAccount and deactivateaccount.
fix https://github.com/acmesh-official/acme.sh/issues/5138
_contains matches with grep regex, so the '*' in "DNS:*.example.com," never
matched and the subject was appended to the identifiers a second time.
Escape the wildcard before the check, the same way the sed removal already does.
fix https://github.com/acmesh-official/acme.sh/issues/5251