Skip to content

Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 49 high) #1008

Open

Description

@github-actions

The weekly CVE scan found fixable HIGH or CRITICAL vulnerabilities in wazuh/wazuh-certs-generator. This body is rewritten by every scan, and the issue closes itself when a scan finds the pinned digest clean.

The fix is normally the Dependabot bump for this image. If upstream has not released one, the fixed versions below say what to wait for.

wazuh/wazuh-certs-generator:0.0.5

Stacks: scratch, soc. Digest: sha256:1be9f9abfb6a37ac7bd3d6d707e1b43589c2b81b23c01a46992e93832795abf9.

Severity ID Package Installed Fixed in
HIGH CVE-2026-92392 curl-minimal 8.21.0-5.amzn2023.0.2 8.21.0-5.amzn2023.0.3
HIGH CVE-2026-86805 glibc 2.34-231.amzn2023.0.5 2.34-231.amzn2023.0.6
HIGH CVE-2026-86805 glibc-common 2.34-231.amzn2023.0.5 2.34-231.amzn2023.0.6
HIGH CVE-2026-86805 glibc-minimal-langpack 2.34-231.amzn2023.0.5 2.34-231.amzn2023.0.6
HIGH CVE-2026-78659 libcap 2.73-1.amzn2023.0.7 2.73-1.amzn2023.0.8
HIGH CVE-2026-78667 libcap 2.73-1.amzn2023.0.7 2.73-1.amzn2023.0.8
HIGH CVE-2026-94440 libcap 2.73-1.amzn2023.0.7 2.73-1.amzn2023.0.8
HIGH CVE-2026-92392 libcurl-minimal 8.21.0-5.amzn2023.0.2 8.21.0-5.amzn2023.0.3
HIGH CVE-2026-103111 pcre2 10.40-1.amzn2023.0.4 10.40-1.amzn2023.0.5
HIGH CVE-2026-86145 pcre2 10.40-1.amzn2023.0.4 10.40-1.amzn2023.0.5
HIGH CVE-2026-89156 pcre2 10.40-1.amzn2023.0.4 10.40-1.amzn2023.0.5
HIGH CVE-2026-89160 pcre2 10.40-1.amzn2023.0.4 10.40-1.amzn2023.0.5
HIGH CVE-2026-103111 pcre2-syntax 10.40-1.amzn2023.0.4 10.40-1.amzn2023.0.5
HIGH CVE-2026-86145 pcre2-syntax 10.40-1.amzn2023.0.4 10.40-1.amzn2023.0.5
HIGH CVE-2026-89156 pcre2-syntax 10.40-1.amzn2023.0.4 10.40-1.amzn2023.0.5
HIGH CVE-2026-89160 pcre2-syntax 10.40-1.amzn2023.0.4 10.40-1.amzn2023.0.5
HIGH CVE-2026-19553 python3 3.9.25-1.amzn2023.0.9 3.9.25-1.amzn2023.0.11
HIGH CVE-2026-19553 python3-libs 3.9.25-1.amzn2023.0.9 3.9.25-1.amzn2023.0.11
HIGH CVE-2026-103242 python3-rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-84837 python3-rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95519 python3-rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95520 python3-rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-95521 python3-rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-103242 rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-84837 rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95519 rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95520 rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-95521 rpm 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-103242 rpm-build-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-84837 rpm-build-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95519 rpm-build-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95520 rpm-build-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-95521 rpm-build-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-103242 rpm-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-84837 rpm-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95519 rpm-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95520 rpm-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-95521 rpm-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-103242 rpm-sign-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-84837 rpm-sign-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95519 rpm-sign-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.9
HIGH CVE-2026-95520 rpm-sign-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-95521 rpm-sign-libs 4.16.1.3-29.amzn2023.0.8 4.16.1.3-29.amzn2023.0.10
HIGH CVE-2026-119371 sqlite-libs 3.40.0-1.amzn2023.0.8 3.40.0-1.amzn2023.0.9
HIGH CVE-2026-119372 sqlite-libs 3.40.0-1.amzn2023.0.8 3.40.0-1.amzn2023.0.9
HIGH CVE-2026-119373 sqlite-libs 3.40.0-1.amzn2023.0.8 3.40.0-1.amzn2023.0.9
HIGH CVE-2026-119375 sqlite-libs 3.40.0-1.amzn2023.0.8 3.40.0-1.amzn2023.0.9
HIGH CVE-2026-119376 sqlite-libs 3.40.0-1.amzn2023.0.8 3.40.0-1.amzn2023.0.9
HIGH CVE-2026-119378 sqlite-libs 3.40.0-1.amzn2023.0.8 3.40.0-1.amzn2023.0.9

Written by .github/workflows/cve-scan.yml in this run.

Activity

  1. added
    securityA credential, an exposure, or a control that is weaker than recorded
    socstacks/soc on odin (VLAN 30)
    scratchstacks/scratch on diabolos (VLAN 30)
    on Oct 7, 2026
  2. changed the title [-]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 232 high)[/-] [+]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 239 high)[/+] on Oct 10, 2026
  3. Gerrrt commented on Oct 10, 2026

    @Gerrrt
    Owner

    Triage (2026-10-10): waiting on upstream, probably for a long time. No change to the pin.

    There is no fixed image, and none is due. 0.0.4 (built 2025-12-17) is the newest tag; the four tags span 2023-08 to 2025-12. It still resolves to the digest we pin. wazuh-docker's v4.14.9-rc2 still points its generate-indexer-certs.yml at wazuh/wazuh-certs-generator:0.0.4, so the 4.14.9 release won't bring a new one either. The Dockerfile is amazonlinux:2023 plus yum update -y && yum install openssl curl-minimal. All 239 findings are Amazon Linux packages that went stale since that build: openssl (×120 across three packages), python3, expat, libcap, glib2, glibc, curl, gnupg2 and so on. Any rebuild clears most of them. There is nothing newer to scan.

    What can be reached. Nothing, the way we run it. The service sits behind profiles: ["certs"], so make up neither starts nor pulls it. It runs once, by hand, with --rm, before odin's first start or to reissue certificates. It publishes no ports and listens on nothing. During that run it reads certs.yml, fetches the cert tool from Wazuh's package bucket over HTTPS with curl, and calls openssl locally to write keys into wazuh/certs/. The only network exposure is that one outbound download, to a Wazuh URL, for the few seconds the run lasts. Between runs no container exists.

    When to act. When a new wazuh/wazuh-certs-generator tag appears: scan it, bump if it is cleaner. It is in the wazuh/* Dependabot group, so it will arrive on a Wazuh release PR. That bump does not touch CERT_TOOL_VERSION. That stays the wazuh/* major.minor. Until a new tag appears there is no action, and this issue will sit open.

    The issue stays open and closes itself when a scan finds the pinned digest clean.

  4. Gerrrt commented on Oct 10, 2026

    @Gerrrt
    Owner

    This is waiting on upstream. Nothing in the repo needs to change.

    • The pinned digest is current: wazuh/wazuh-certs-generator:0.0.4 (sha256:369b4d58…). pin-digests.sh reports no drift, so the tag hasn't been rebuilt since it was pinned.
    • There is no newer release on the registry to move to.
    • The fixed versions exist (table above), but no published image includes them yet. Mostly:
      • openssl: 40 findings, fixed in 6 different versions (see the table)
      • openssl-fips-provider-latest: 40 findings, fixed in 6 different versions (see the table)
      • openssl-libs: 40 findings, fixed in 6 different versions (see the table)
      • python3: 19 findings, fixed in 5 different versions (see the table)
      • python3-libs: 19 findings, fixed in 5 different versions (see the table)
      • expat: 15 findings, fixed in 3 different versions (see the table)
      • and 66 more across 26 other packages

    When upstream rebuilds or ships a release with these fixes, Dependabot should open the bump. Check it with trivy and check_image_pins.py, then deploy. The next scan will then close this issue.

  5. Gerrrt commented on Oct 10, 2026

    @Gerrrt
    Owner

    Rechecked 2026-10-10. One change to the triage above: a fixed wazuh-certs-generator image is unlikely ever to arrive, because Wazuh 5.0 drops the image.

    • Docker Hub still has only 4 tags. 0.0.4 was last pushed 2025-12-17 and is still our digest.
    • In wazuh-docker v5.0.0-rc1 (2026-10-05), single-node/docker-compose.yml has no cert-generator service and no generate-indexer-certs.yml. Its deployment docs download https://packages.wazuh.com/5.0/wazuh-certs-tool-5.0.0-1.sh and run it on the host through tools/utils/deployment/certificates-conf.sh --cert --copy --priv. Nothing in that tree references wazuh/wazuh-certs-generator.
    • The 4.14.x line (v4.14.9-rc2) still uses 0.0.4, so staying on 4.14 means staying on this image.

    So "wait for a new tag" probably never closes this. It closes when the SOC stack moves to Wazuh 5.0 and the certs service is removed from stacks/soc/compose.yaml, along with the copy in stacks/scratch. The scan then closes it as no longer pinned. Two follow-ups:

    • The 5.0 move needs its own issue. It is a major upgrade of the SIEM, cert issuance moves to a host-side script, and the manager gains a remoted cert (wazuh.manager-remoted.pem). Not in scope here.
    • Until then the exposure assessment above stands: this runs once, by hand, with --rm, behind profiles: ["certs"], and listens on nothing. No action on this issue in the meantime.

    scripts/prune_superseded_images.py also names this image in its tests, so check it when the service goes.

  6. Gerrrt commented on Oct 10, 2026

    @Gerrrt
    Owner

    What the generator actually runs, and what that exposes, 2026-10-10

    The triage above says nothing is reachable "the way we run it", and calls the one outbound download the only network exposure. This traces that run step by step against the pinned wazuh-certs-generator:0.0.4@sha256:369b4d58…. All 239 rows are covered (121 distinct CVEs). It also turns up something more important than any of them: the run sources an unverified script from the network, as root, with our certificate directory mounted read-write.

    What a run does (/entrypoint.sh in the image, plus the tool it downloads):

    1. curl -fsL https://packages.wazuh.com/4.14/wazuh-certs-tool.sh. If that fails for any reason (a 404, a timeout, a 5xx), it tries https://packages-dev.wazuh.com/4.14/ instead, and says so only in a log line.
    2. source /wazuh-certs-tool.sh -A: no checksum, no signature, as root.
    3. The tool's -A path calls openssl genrsa and openssl req, and parses our certs.yml with grep, sed and awk. It defines helpers that call rpm, yum, gpg and curl, but nothing on any path calls them.
    4. The entrypoint copies the results into /certificates (stacks/soc/wazuh/certs/), including the CA private key twice: root-ca.key and root-ca-manager.key, the second owned by uid 999.

    The 239 rows against that run:

    Packages Rows Verdict
    Network-facing, for seconds per manual run OpenSSL as curl's TLS client; curl-minimal, libcurl-minimal, libnghttp2 part of OpenSSL's 120, plus 3 Reachable only by something on the path to packages.wazuh.com (or packages-dev) while the fetch runs. I haven't traced the 40 OpenSSL CVEs one by one. See below for why it matters less than it looks.
    On the path, local input only OpenSSL as the openssl CLI; gawk; pcre2 (grep, sed); libacl; glibc; libtasn1 rest of the 120, plus 11 Their input is our certs.yml, the image's own CA bundle, and keys the run generates. Not attacker-controlled.
    Never executed python3, python3-libs, python3-pip-wheel, expat (via Python), libcap, glib2, libsolv, rpm ×5 packages, sqlite-libs, gnupg2-minimal, libblkid/libmount/libsmartcols/libuuid 105 Nothing in the entrypoint or in the tool's -A path runs them.

    Why the unverified download matters more than the CVEs. Anyone who could exploit a pre-verification TLS bug during that fetch gets code running in the generator, with the CA key in reach. Anyone who can change what packages.wazuh.com/4.14/wazuh-certs-tool.sh serves gets the same thing, with no bug needed. And that file does change under the same URL: it was last modified 2026-09-23 (sha256:8c93ed36d7b956a6e97a906aed6b6bc636d7cb55a15a41fd6e9c2aa825164216 today). Whatever is there when the job runs generates our CA. The silent fallback to the dev bucket widens that to a second bucket, which answers 200 today. So patching OpenSSL wouldn't close the real exposure. Pinning the tool would: run it only if its hash matches a value committed here, and refuse the dev bucket. That is a change to the certs service, not to this issue, and it's flagged separately.

    What doesn't change. It still runs only by hand, behind profiles: ["certs"], with --rm, and nothing listens. The 15:53 conclusion stands: no fixed image is coming, and this closes when the SOC moves to Wazuh 5.0, which drops the image.

    Aside on the CA key. The runbook's §6 says root-ca.key "stays in that gitignored directory and nowhere else". That holds, but the directory holds two copies, root-ca.key and root-ca-manager.key. Neither is mounted into a container: compose mounts only root-ca-manager.pem.

    To reproduce:

    IMG=wazuh/wazuh-certs-generator:0.0.4@sha256:369b4d58509aab074b188596870c81584f7120e653d9ef83c591f0f785dcf325
    docker run --rm --network none --entrypoint cat "$IMG" /entrypoint.sh
    curl -fsSI https://packages.wazuh.com/4.14/wazuh-certs-tool.sh | grep -i last-modified
    curl -fsS https://packages.wazuh.com/4.14/wazuh-certs-tool.sh | sha256sum

    🤖 Traced by Claude Code

  7. Gerrrt commented on Oct 11, 2026

    @Gerrrt
    Owner

    Update 2026-10-11: the exposure the 22:33 trace flagged is closed by #1168 (merged 04:25 UTC).

    • The certs service in soc and scratch no longer runs the image's entrypoint. It runs scripts/wazuh-certs-mint.sh, which uses a vendored wazuh-certs-tool.sh (sha256 8c93ed36…, recorded in stacks/soc/wazuh/certs-tool/VENDORED) and refuses to run if the hash doesn't match. There's no download, no fallback to packages-dev, and nothing unverified is sourced.
    • The service now runs with network_mode: none. feat(soc): mint the SOC's CA offline with a cert tool pinned by sha256 #1168's proof showed curl can't resolve hosts inside it, and only lo exists. validate.sh runs vendor-wazuh-certs-tool.sh --check.

    What that does to the 239 rows: the 22:33 table's one network-facing group (OpenSSL as curl's TLS client, plus curl-minimal, libcurl-minimal and libnghttp2) is gone, because curl no longer runs and there's no network. Every row is now either local input only (openssl CLI on keys the run generates, plus grep/sed/awk over our certs.yml) or never executed. None is reachable.

    What doesn't change: the image is still pinned (0.0.4@sha256:369b4d58…, used for its bash and openssl userland), so the scan still lists the 239 rows and this issue stays open. It still closes when the SOC moves to Wazuh 5.0 and the certs service goes, or if a newer generator tag ever appears. When that happens, scripts/wazuh-certs-mint.sh, the vendored tool and prune_superseded_images.py's test fixture all need updating too.

  8. Gerrrt commented on Oct 11, 2026

    @Gerrrt
    Owner

    Tracked in #1176: this closes when the SOC moves to Wazuh 5.0, which drops the image that carries these findings. Blocked until 5.0 is GA.

  9. changed the title [-]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 239 high)[/-] [+]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 266 high)[/+] on Oct 19, 2026
  10. changed the title [-]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 266 high)[/-] [+]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 288 high)[/+] on Oct 26, 2026
  11. changed the title [-]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 288 high)[/-] [+]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 285 high)[/+] on Nov 2, 2026
  12. changed the title [-]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 285 high)[/-] [+]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 15 high)[/+] on Nov 9, 2026
  13. changed the title [-]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 15 high)[/-] [+]Fixable CVEs in wazuh/wazuh-certs-generator (0 critical, 49 high)[/+] on Nov 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    scratchstacks/scratch on diabolos (VLAN 30)securityA credential, an exposure, or a control that is weaker than recordedsocstacks/soc on odin (VLAN 30)

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions