← Back to Feed CACHED · 2026-08-20 10:46:17 · CACHE_KEY CVE-2026-73570
CVE-2026-73570 · CWE-78 · Disclosed 2026-08-13

A remote code execution vulnerability exists in Zimbra Collaboration

ASSESSED — NOISGATE V0.5
Vendor
Reassessed
Verdict:
Do you agree?
01 · The Real Story

Someone left the SNMP door unlocked on the mailroom, and burglars are already inside

CVE-2026-73570 is an unauthenticated OS command injection (CWE-78) in Zimbra Collaboration Suite's SNMP notification processing path. When the optional zimbra-snmp package is installed and the snmp_notify parameter is active — with the swatchdog watchdog service running (which is on by default) — an attacker can send specially crafted SMTP-layer requests that escape input boundaries and execute arbitrary shell commands as the zimbra user. All ZCS versions before 10.1.20 are affected. The patch shipped July 20, 2026; the CVE was published August 13, 2026.

Zimbra scored this HIGH at 8.9 with AC:H (high attack complexity). That rating is defensible on paper — the optional package dependency genuinely narrows the vulnerable population. But in practice, CERT Polska issued Alert 145/2026 on August 17 confirming active exploitation in the wild, Shadowserver tracks over 12,100 internet-exposed Zimbra instances, and the zimbra user has sweeping access to mail spools, application directories, and service configs. For organizations that have zimbra-snmp installed, the vendor's HIGH is accurate; we agree and hold at 8.9. The optional-package friction is the only thing keeping this from CRITICAL.

"Pre-auth RCE on internet-facing mail servers with active exploitation — patch or disable SNMP now."
02 · The Attack Path

5 steps from start to impact.

STEP 01

Identify exposed Zimbra instance

The attacker scans for internet-facing Zimbra servers using Shodan, Censys, or FOFA banner matching on ports 443/7071. Shadowserver currently tracks ~12,100 exposed instances worldwide. The attacker fingerprints the ZCS version via response headers or the login page.
Conditions required:
  • Target Zimbra instance is internet-facing on HTTPS
Where this breaks in practice:
  • Organizations using Zimbra behind a VPN or reverse proxy with no public exposure are unreachable
Detection/coverage: Shodan/Censys queries for zimbra banner; GreyNoise may tag scanning activity
STEP 02

Confirm zimbra-snmp is active

The attacker probes for indicators that the zimbra-snmp optional package is installed and SNMP notifications are enabled (snmp_notify parameter). The swatchdog service runs by default when the package is present, so no additional admin action is needed beyond having installed the package.
Conditions required:
  • zimbra-snmp package installed on target
  • snmp_notify parameter enabled
  • swatchdog service running (default when package present)
Where this breaks in practice:
  • zimbra-snmp is explicitly optional — many deployments skip it entirely
  • Administrators who never configured SNMP monitoring are not vulnerable
STEP 03

Deliver command injection payload

The attacker crafts a specially formed SMTP request that reaches the SNMP notification processing code path. Due to improper sanitization (CWE-78), shell metacharacters in the payload escape input boundaries and are passed to underlying system calls. No authentication is required. The CVSS vector marks this AC:H, suggesting the injection point requires specific timing or payload construction.
Conditions required:
  • Network access to the Zimbra SMTP/HTTPS interface
  • Knowledge of the injection syntax (no public PoC code, but attackers have it)
Where this breaks in practice:
  • AC:H in the CVSS vector implies non-trivial exploitation — not a single curl command
  • No public proof-of-concept code is available as of August 2026
Detection/coverage: WAF rules inspecting SMTP payloads for shell metacharacters; Zimbra log anomalies in SNMP-related entries
STEP 04

Achieve command execution as zimbra user

Successful injection executes arbitrary OS commands under the zimbra service account. This account has read/write access to all mailbox data under /opt/zimbra/store/, Jetty web application directories, configuration files, and can restart services. The attacker typically deploys a JSP web shell to /opt/zimbra/jetty/webapps/ or /opt/zimbra/jetty_base/webapps/ for persistent access.
Conditions required:
  • Successful command injection from step 3
Where this breaks in practice:
  • The zimbra user is not root — privilege escalation to root requires a separate chain
  • File integrity monitoring on webapps directories would flag web shell drops
Detection/coverage: CERT Polska IOCs: check for unexpected files in /opt/zimbra/jetty/webapps/, /opt/zimbra/jetty_base/webapps/, and /tmp/ created by the zimbra user within the last 30 days; unexpected Zimbra service restarts
STEP 05

Data exfiltration and lateral movement

With persistent access as the zimbra user, the attacker can read all email for all users on the server, harvest stored credentials from LDAP configurations, modify mail routing rules, and use the compromised mail server as a pivot point for internal network reconnaissance. State-backed actors have historically used Zimbra compromises for long-term espionage.
Conditions required:
  • Persistent web shell or reverse shell from step 4
Where this breaks in practice:
  • Network segmentation may limit lateral movement from the mail server
  • DLP and egress monitoring can detect bulk mail exfiltration
Detection/coverage: Monitor outbound traffic from Zimbra servers for unusual volumes or destinations; SIEM correlation on zimbra user process spawning unexpected child processes (curl, wget, nc, python)
03 · Intelligence Metadata

The supporting signals.

In-the-Wild ExploitationActive. CERT Polska Alert 145/2026 (August 17, 2026) confirms exploitation. No specific threat actor attributed, but Zimbra has been repeatedly targeted by Russian state-backed groups (GRU/APT28) in prior campaigns.
Proof of ConceptNo public PoC as of 2026-08-20. Attackers have working exploits per CERT Polska confirmation, but no code has surfaced on GitHub, Exploit-DB, or researcher blogs. This limits opportunistic mass exploitation — for now.
EPSS0.00539 (≈95th percentile). Low probability score, but EPSS historically lags behind active exploitation signals for newly disclosed vulns.
KEV StatusNot listed as of 2026-08-20. Given active exploitation confirmation from CERT Polska, KEV addition is likely imminent.
CVSS VectorCVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:C/C:H/I:H/A:L — Network-accessible, no auth, no user interaction, but high attack complexity. Scope change means the zimbra user can impact resources beyond the vulnerable component (e.g., other mailboxes, other services on the host).
Affected VersionsAll Zimbra Collaboration Suite releases before 10.1.20. Both Network Edition and Open Source Edition are affected when zimbra-snmp is installed.
Fixed Version10.1.20 (released July 20, 2026). No known distro backports — Zimbra is typically self-hosted from vendor packages.
Exposure SurfaceShadowserver tracks ~12,100 internet-exposed Zimbra servers globally, concentrated in Europe (~3,800) and Asia (~3,800). Not all have zimbra-snmp installed.
Disclosure TimelinePatch released 2026-07-20 → CVE published 2026-08-13 → CERT Polska active exploitation alert 2026-08-17.
Reporting EntityZimbra internal discovery; CERT Polska (CSIRT for Poland) issued the active exploitation warning.
04 · The Call

noisgate verdict.

Final Verdict
= UNCHANGED to HIGH (8.9/10)

Active in-the-wild exploitation of a pre-auth RCE on internet-facing mail servers is the decisive factor sustaining the HIGH rating. The optional zimbra-snmp package dependency is the single strongest friction point preventing escalation to CRITICAL — it materially narrows the vulnerable population below the threshold where every Zimbra instance is at risk.

HIGH Vulnerability existence and active exploitation
MEDIUM Fraction of Zimbra deployments with zimbra-snmp installed
LOW Specific threat actor attribution and campaign scope

Why this verdict

  • Active exploitation confirmed: CERT Polska Alert 145/2026 documents real-world attacks as of August 17, 2026. This is not theoretical.
  • Pre-auth RCE on internet-facing infrastructure: Zimbra mail servers are internet-exposed by design. No credentials or user interaction required. The zimbra user has access to all organizational email — this is a mass-data-exfiltration outcome.
  • Role multiplier: Zimbra is a mail server, canonically internet-facing and holding all organizational email. In environments where Zimbra also serves as the LDAP directory (common in smaller orgs), compromise extends to identity-store access. Blast radius: host → all mailboxes → potential credential harvest → lateral movement. This satisfies the HIGH floor for a high-value-role component.
  • Optional package narrows population: zimbra-snmp is not installed by default — it is an explicit opt-in during deployment. This means a meaningful fraction of the ~12,100 exposed instances are *not* vulnerable even if unpatched. This is the primary reason the verdict stays HIGH rather than escalating to CRITICAL.
  • AC:H limits mass exploitation: The CVSS vector's high attack complexity suggests the injection requires specific payload construction, not a trivial single-request exploit. Combined with no public PoC, this raises the attacker skill floor.

Why not higher?

CRITICAL would require that the vulnerability affects all or nearly all Zimbra deployments without additional prerequisites. The zimbra-snmp package is optional and must be explicitly installed — we lack data on adoption rates but it is reasonable that a significant fraction of deployments (potentially 30-60%) never install SNMP monitoring. Additionally, AC:H indicates non-trivial exploitation, and no public PoC exists to enable mass commodity attacks. These two factors together keep the effective population and ease-of-exploitation below the CRITICAL threshold.

Why not lower?

Active exploitation in the wild immediately disqualifies any downgrade below HIGH. The attack is pre-auth, network-accessible, and targets a component (mail server) that is internet-facing by design and holds highly sensitive data. The zimbra user's access to all mailboxes means successful exploitation has organization-wide data impact. Even with the optional package friction, confirmed ITW exploitation means the chain is proven viable and attackers have already solved the complexity barrier.

05 · Compensating Control

What to do — in priority order.

  1. Disable SNMP notifications immediately — Run zmprov ms $(zmhostname) zimbraSnmpNotificationEnabled FALSE and restart the swatchdog service. This eliminates the vulnerable code path entirely without disrupting mail delivery. Deploy within hours given active exploitation.
  2. Block SNMP-related ports at the perimeter — If you cannot disable SNMP notifications at the application layer, ensure UDP 161/162 and any custom SNMP trap ports are not reachable from untrusted networks. Note: the injection vector may traverse SMTP/HTTPS, so this alone is insufficient.
  3. Audit for web shells on Zimbra servers — Immediately search /opt/zimbra/jetty/webapps/, /opt/zimbra/jetty_base/webapps/, and /tmp/ for JSP or other script files created by the zimbra user in the last 30 days. Cross-reference CERT Polska IOCs from Alert 145/2026. Treat any finding as confirmed compromise.
  4. Deploy WAF rules for SMTP command injection patterns — If Zimbra is behind a WAF or reverse proxy, add rules to inspect and block SMTP payloads containing shell metacharacters (;, |, $(), backticks). This is a partial control — effectiveness depends on the specific injection syntax attackers are using.
  5. Restrict outbound network access from Zimbra servers — Limit egress from Zimbra hosts to only required destinations (DNS, patch repos, upstream mail relays). This constrains post-exploitation data exfiltration and C2 communication. Deploy within 3 days.
What doesn't work
  • Network segmentation alone does not help — Zimbra mail servers must be internet-accessible to function, so the attack surface cannot be eliminated through network controls.
  • Disabling the zimbra user is not feasible — the entire Zimbra application runs under this account. Restricting it breaks mail service.
  • SNMP ACLs on the Zimbra host may not help because the injection vector appears to traverse SMTP/HTTPS request processing, not direct SNMP protocol access to ports 161/162.
06 · Verification

Crowdsourced verification payload.

Run this script on each Zimbra server as root or the zimbra user. It checks whether the zimbra-snmp package is installed, whether SNMP notifications are enabled, and whether the installed version is vulnerable. Example: sudo bash zimbra_cve_2026_73570_check.sh

noisgate-verify.sh
BASHREAD-ONLYSAFE
#!/bin/bash
# CVE-2026-73570 Vulnerability Check for Zimbra Collaboration
# Run on each Zimbra server as root or zimbra user
# Exit codes: 0=PATCHED, 1=VULNERABLE, 2=UNKNOWN

ZIMBRA_BIN="/opt/zimbra/bin"
VULN_STATUS="UNKNOWN"

# Check if Zimbra is installed
if [ ! -d "$ZIMBRA_BIN" ]; then
  echo "UNKNOWN - Zimbra not found at /opt/zimbra"
  exit 2
fi

# Get Zimbra version
VERSION=$($ZIMBRA_BIN/zmcontrol -v 2>/dev/null | grep -oP '[0-9]+\.[0-9]+\.[0-9]+')
if [ -z "$VERSION" ]; then
  echo "UNKNOWN - Could not determine Zimbra version"
  exit 2
fi

echo "Zimbra version detected: $VERSION"

# Check if zimbra-snmp package is installed
SNMP_INSTALLED=false
if dpkg -l zimbra-snmp 2>/dev/null | grep -q '^ii'; then
  SNMP_INSTALLED=true
  echo "zimbra-snmp package: INSTALLED (via dpkg)"
elif rpm -q zimbra-snmp 2>/dev/null | grep -qv 'not installed'; then
  SNMP_INSTALLED=true
  echo "zimbra-snmp package: INSTALLED (via rpm)"
else
  echo "zimbra-snmp package: NOT INSTALLED"
  echo "PATCHED - zimbra-snmp not installed, CVE-2026-73570 does not apply"
  exit 0
fi

# Check if SNMP notifications are enabled
SNMP_ENABLED=$($ZIMBRA_BIN/zmprov gs $($ZIMBRA_BIN/zmhostname) zimbraSnmpNotificationEnabled 2>/dev/null | grep -i 'TRUE')
if [ -n "$SNMP_ENABLED" ]; then
  echo "SNMP notifications: ENABLED (vulnerable code path active)"
else
  echo "SNMP notifications: DISABLED (vulnerable code path inactive)"
  echo "PATCHED - SNMP notifications disabled, attack vector mitigated"
  exit 0
fi

# Compare version - vulnerable if < 10.1.20
MAJOR=$(echo "$VERSION" | cut -d. -f1)
MINOR=$(echo "$VERSION" | cut -d. -f2)
PATCH=$(echo "$VERSION" | cut -d. -f3)

if [ "$MAJOR" -lt 10 ]; then
  VULN_STATUS="VULNERABLE"
elif [ "$MAJOR" -eq 10 ] && [ "$MINOR" -lt 1 ]; then
  VULN_STATUS="VULNERABLE"
elif [ "$MAJOR" -eq 10 ] && [ "$MINOR" -eq 1 ] && [ "$PATCH" -lt 20 ]; then
  VULN_STATUS="VULNERABLE"
else
  VULN_STATUS="PATCHED"
fi

echo ""
if [ "$VULN_STATUS" = "VULNERABLE" ]; then
  echo "VULNERABLE - Zimbra $VERSION with zimbra-snmp and SNMP notifications enabled"
  echo "  -> Upgrade to 10.1.20+ or disable SNMP notifications immediately"
  echo "  -> Check for IOCs: find /opt/zimbra/jetty/webapps /opt/zimbra/jetty_base/webapps /tmp -user zimbra -mtime -30 -type f"
  exit 1
else
  echo "PATCHED - Zimbra $VERSION is at or above 10.1.20"
  exit 0
fi
07 · Bottom Line

If you remember one thing.

TL;DR
This is an actively exploited pre-auth RCE on mail servers — treat it as an emergency regardless of the HIGH label. Because CERT Polska confirmed in-the-wild exploitation on August 17, the noisgate active-exploitation override applies: mitigate immediately, within hours, by disabling SNMP notifications (zimbraSnmpNotificationEnabled FALSE) on every Zimbra server. Simultaneously, begin upgrading to ZCS 10.1.20 under the noisgate remediation SLA for HIGH (≤180 days), though you should aim to complete patching within 2 weeks given active exploitation. Before patching, sweep every Zimbra host for web shells in /opt/zimbra/jetty/webapps/ and /tmp/ per CERT Polska's IOC guidance — any findings mean you have a confirmed breach, not just a vulnerability. If you manage 10,000 hosts and Zimbra is your mail platform, this is your #1 priority Monday morning.

Sources

  1. NVD - CVE-2026-73570
  2. CERT Polska Alert / OffSeq Threat Radar
  3. Zimbra Security Center
  4. Zimbra 10.1.20 Patch Release Blog
  5. CyberSecurityNews - Zimbra SNMP Command Injection
  6. GBHackers - Zimbra RCE Vulnerability
  7. CyberPress - Zimbra OS Command Injection Exploited
  8. Canadian Centre for Cyber Security Advisory
Peer Review

What defenders are saying.

Submit a review attribution: handle + country only
0 flags selected · stored anonymously
Validation Results

Crowdsourced verification outputs.

Results submitted by users who ran the verification payload against their environment.