Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

esxcli software profile update fails: [DependencyError], [MissingComponentError] or 'Could not find a trusted signer'

Technical article
VMware

About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.

When esxcli software profile update stops with [DependencyError], [MissingComponentError] or [InstallationError] with "Could not find a trusted signer", check the host's state before you assume anything: run esxcli software profile get and esxcli software vib list and compare them with what you expect. Each error points at a different mismatch. A dependency error means a VIB already on the host, for example from the server maker's add-on, requires something the profile you chose cannot provide. A missing component error is documented by Broadcom for a host whose image came from vSphere Lifecycle Manager, where the command line cannot reconcile it. A trusted-signer error means the host does not trust the authority that signed a VIB.

Before you try again, run the same command with --dry-run. It reports what would happen without changing anything.

What the error means

The command line works with VIBs (individual software packages), components and image profiles (a named set of VIBs and components). Three commands are easy to mix up.

  • esxcli software profile update updates the host with the VIBs from an image profile. The reference says installed VIBs may be upgraded but will not be removed, and that VIBs in the profile unrelated to installed ones are added.
  • esxcli software profile install replaces the installed image with the profile and may result in the loss of installed VIBs.
  • esxcli software vib install and update work on single VIBs. Broadcom says plainly that they "are not supported for upgrade operations."

The dependency errors have a recognisable shape: VIB <name> requires <something>, but the requirement cannot be satisfied within the ImageProfile. Broadcom's examples include esx-base requiring esx-update at a matching level, a QLogic driver requiring qedentv_ver, and, for an 8.x to 9.x attempt with a vendor image, Mellanox VIBs requiring esx-version < 8.1.0.

The signature error comes from checking each VIB's signature against the host's trust store. Broadcom's article on third-party VIBs says that when the signing root or intermediate authority is not in the host's trusted store, validation fails with VibSigInvalidError "even if the VIB itself is not corrupted".

Likely causes

Cause What you see How to confirm
vib update or vib install used for an upgrade [DependencyError] esx-base requires esx-update Use profile update with -p and the bundle
Wrong profile for an OEM-customised host "requires ..., but the requirement cannot be satisfied within the ImageProfile" esxcli software profile get shows an OEM profile name
OEM VIB tied to an older ESXi version requires esx-version < 8.1.0 against a 9.0 profile The VIB name and its version limit in the message
Host image built by Lifecycle Manager [MissingComponentError] "Missing reserved components ..." esxcli software profile get shows "(Updated) VMware Lifecycle Manager Generated Image"
Old host, new signing certificate [InstallationError] "Could not find a trusted signer." upgrading from an old build Host runs 6.0 GA up to before build 9239799, or 6.5 GA up to before build 8294253 (Broadcom's failing range)
Third-party VIB with an untrusted signing authority "Could not find a trusted signer: unable to get local issuer certificate" esxupdate or lifecycle.log shows VibSigInvalidError
Certificate or time trouble "Could not find a trusted signer: self-signed certificate" vCenter certificates expired, host time wrong

Prove the cause before changing anything

Read-only checks on the host, from an SSH session.

esxcli software profile get
esxcli software vib list
esxcli software acceptance get
esxcli software sources profile list -d <offline bundle .zip path>
esxcli software profile update -d <offline bundle .zip path> -p <profile name> --dry-run

profile get shows how the host was built. Broadcom contrasts two forms: a host installed or upgraded by Lifecycle Manager shows "(Updated) VMware Lifecycle Manager Generated Image", while one installed manually or upgraded with esxcli shows a name such as HPE-Custom-AddOn_802.0.0.11.6.0-5. sources profile list lists the profile names in the bundle, so you choose one, not guess. The reference defines --dry-run for profile update as reporting the VIB-level operations that would be performed without changing anything. The log for failures is /var/log/esxupdate.log (named on Broadcom's page 318791, and its examples show tracebacks there) and lifecycle.log for Lifecycle Manager.

Check whether a suspect VIB's driver is in use before you consider removing it. Broadcom's check is esxcfg-module -l | grep -i <module-name>; no output means the module is not loaded. esxcfg-scsidevs -a and esxcfg-nics -l show the storage and network devices the drivers serve.

Example (illustrative): esx01 runs an 8.x image from the server maker. The target is the 9.0 profile from the maker's depot. profile update stops with [DependencyError] on platform embeddedEsx: a Mellanox VIB requires esx-version < 8.1.0. The target is version 9.0, which is not below 8.1.0, so no version of 9.0 can satisfy it. Removing the VIB is a decision for the hardware vendor, because the VIB may be part of the vendor's supported image.

Fix

Put the host in maintenance mode with its VMs moved away first, as Broadcom's procedure says. The reference also warns that if an installation requires a reboot, you need to disable HA first.

Risk: disabling HA means the cluster will not restart VMs after a host failure while it is off. Re-enable HA as soon as the host is back from its reboot, and keep the window short.

Running the profile update

Use this sequence each time the update is run, once the cause of an error has been dealt with.

  1. Save the starting point. Run esxcli software profile get and esxcli software vib list, and keep both outputs where they will survive the host being out of service. They record what to return to.
  2. Run the update with --dry-run first: esxcli software profile update -d <bundle> -p <profile name> --dry-run. This is the command's own check, and the reference describes it as reporting the VIB-level operations that would be performed without changing anything. Stop if it reports an error.
  3. Disruptive step: run esxcli software profile update -d <bundle> -p <profile name> without --dry-run. The update changes the installed image and needs a reboot to take effect, so the host is out of service while it restarts. Run it only on a host in maintenance mode with its VMs moved away, and with HA handled as in the Risk above. There is no one-command undo. To back out, update again with the bundle and profile for the previous version, using the outputs from step 1 to name it, and add --allow-downgrades: the ESXCLI reference says installed VIBs "may be upgraded (or downgraded if --allow-downgrades is specified)", and without the option only VIBs that update or are new to the host are installed.
  4. Reboot the host if the command's output says a reboot is required, then check with esxcli software profile get and esxcli software vib list against the saved outputs.

Dependency errors

  1. Use the right command: list the profiles with esxcli software sources profile list -d <bundle>, then run the update as in the sequence above. Broadcom's esx-base example is fixed this way. KB 316421 also prints a vib install method but ends by saying that vib install and vib update are not supported for upgrade operations, so the steps here follow the closing statement and the profile method.
  2. For OEM images, use the maker's bundle that matches the target version.
  3. If a single VIB still blocks, find out whether it is in use (the esxcfg-module check). Then remove it with esxcli software vib remove -n <vib-name> and reboot before retrying the upgrade.

Disruptive step: removing a VIB can remove a driver or management agent the host needs, and the reboot takes the host out of service. Do it only on a host in maintenance mode, with the maker's agreement for OEM VIBs. Broadcom's 8.x to 9.x case says to consult the vendor first. To back out, reinstall the VIB from the vendor's bundle.

MissingComponentError on a host managed by Lifecycle Manager

Broadcom's workaround is not to use the command line on that host. Patch it with vSphere Lifecycle Manager instead. Removing a named VIB does not help, as Broadcom shows the same error still naming it afterwards.

Could not find a trusted signer

  1. Old host builds. Broadcom explains that a new signing certificate was introduced into the trust store in 2018 and the old certificate expired on 31 December 2019, so releases from before the transition do not trust the new signatures. Upgrade the host to at least ESXi600-201807001 (build 9239799) for 6.0, 6.5 U2 (build 8294253) or 6.7 GA (build 8169922) first.
  2. Certificates and time. Broadcom's NSX article lists expired vCenter certificates, a host clock well out from vCenter, a damaged trust store file /etc/vmware/ssl/castore.pem and an expired transport node certificate. Check host time using the guide "ESXi host time is wrong: NTP shows 'Time Synchronized: false' or 'Configuration is not working normally'".
  3. Third-party VIB. Broadcom's first option is to get a VIB from the vendor signed by a VMware-trusted authority.

Broadcom's second option is a bypass. If the VIB is categorised as CommunitySupported, Broadcom's page lowers the host's acceptance level with esxcli software acceptance set --level CommunitySupported (the reference lists VMwareCertified, VMwareAccepted, PartnerSupported and CommunitySupported); that step applies only in that case. The page then installs with --no-sig-check.

Risk: the reference says --no-sig-check "poses a large security risk" and shows a SECURITY ALERT warning in the vSphere Client. The host becomes Non-Compliant in Lifecycle Manager until a signed version is installed. Secure Boot is designed to disallow unsigned VIBs and also disallows the --force flag. The reference says --force should be used only when VMware Technical Support instructs. To back out, set the acceptance level back to its previous value (read it first with esxcli software acceptance get) and install a properly signed VIB.

Confirm the fix

  • profile update ends with a report of VIBs installed, removed and skipped, and whether a reboot is required.
  • After the reboot, esxcli software profile get shows the new profile and esxcli software vib list shows the expected versions.
  • The host reconnects in vCenter and passes a Lifecycle Manager compliance check, if you use one.

What to watch afterwards

  • Run --dry-run for every future command-line update.
  • Keep the maker's bundle and Broadcom's bundle versions in step.
  • Decide per host whether command-line patching or Lifecycle Manager owns it. Do not alternate.
  • Record any acceptance level changed and put it back.

Risks and limits

  • These commands have not been run or lab-tested. They come from Broadcom's knowledge base articles and the ESXCLI reference.
  • The messages quoted here are the ones Broadcom prints.
  • The reference describes --dry-run; it does not say whether signature checks run during a dry run. A dry run that succeeds does not prove the real run will.
  • --force and --no-sig-check weaken the host's protections. Broadcom's reference limits --force to use "only when instructed to do so by VMware Technical Support".
  • Broadcom's third-party VIB and NSX articles are about specific cases. A site with a different signing issue should raise a support request.

Sources