Skip to content

vSphere Integrated Containers Engine Version v1.4.1
Compare
Choose a tag to compare
@zjs zjs released this 03 Jul 15:05
v1.4.1
b3a6192

What's in the Release Notes

Download Binaries

Installation and Upgrade

Using vSphere Integrated Containers

For more details on using vSphere Integrated Containers see the end user documentation at https://vmware.github.io/vic-product/#documentation.

Changes

Full list of changes from 1.4.0.

New Features

This release includes the enhancements and bug fixes listed below.

Enhancements

This release includes the following enhancements to existing features:

  • Added the ability to manage use of DRS VM affinity groups for existing VCHs using vic-machine configure.

Resolved Issues

The following issues that were documented as known issues in previous releases or reported by customers have been fixed in v1.4.1:

  • docker exec always returns 0 and ignores the exit code of processes. #5692
    docker exec always returns 0, even if you specify -it. This issue is potentially due to a delay in vSphere host synchronization. If you configure command-based health checks in vSphere Integrated Containers Management Portal, the health checks are always successful for containers that are provisioned on affected VCHs, even if the user-specified command does not exist in the scope of the container. This is because command-based health checks are considered to be successful if the exit code of that command is 0.
  • Concurrent docker exec on the same container causes conflict errors. #7410 (mitigation, not full fix)
    Concurrent dispatch of docker exec returns conflict errors. These errors are expected when multiple operations attempt to modify a containerVM at once; at least one should succeed, with the group as a whole making progress until all are cleanly dispatched. A combination of low level behaviours prevent that consistent progress from occurring, some of which are still under investigation or require work beyond the scope of a patch release to address.
    The 1.4.1 patch release includes some fixes and a change that serializes the initial dispatch of exec against the same container to mitigate this problem. It is only the dispatch of exec that is serialized, not the execution of them. Testing has been done up to 30 concurrent execs, but it's highly recommended to keep the number low. The time to complete a given number of execs is highly dependent on infrastructure and load, and some clients have been observed to hit HTTP connection timeouts if blocked for long during this operation.

Additionally, the following other issues have been addressed:

  • Manually-configured DNS servers now override DHCP-provided DNS servers. #7824
    Previously, DNS servers configured on a VCH were combined with DNS servers provided via DHCP.
  • Increased the maximum number of memory maps a container may use. #7790
    Previously, the maximum number of memory maps was limited to 65530. This was insufficient for some workloads (such as Elasticsearch).

See also the resolved issues for each of the other vSphere Integrated Containers components:

Known Issues

With the exception of issues resolved above, and the new known issue below, v1.4.1 has the same known issues as v1.4.0.

  • NFS volume stores do not support FQDN. #8043
    When using FQDN to identify NFS shares, it is not possible to set a DNS-Search-List. Container VMs fail to start, with the error docker: Error response from daemon: Server error from portlayer: unable to wait for process launch status: container VM has unexpectedly powered off.

    Workaround: Always use IP addresses for NFS shares.

View all

  • Create Virtual Container Host wizard plugin throws a parsing error when you specify only the gateway for the management or client networks. #7602
    When you provide a gateway for the management or client network, it is mandatory to provide at least one routing destination. The Create Virtual Container Host wizard does not conform to this behavior and has two separate fields for the Gateway and Routing destinations. If you do not fill in both the fields correctly, the wizard reports a parsing error.

  • Attempts to change operations user permissions with vic-machine configure do not roll back in the event of a failure. #7814
    Changes to the operations user that are made during a failed vic-machine configure operation persist even after a rollback.

    Workaround: In the vSphere Client, go to Administration -> Roles and delete every role that begins with vch. WARNING: Only perform this workaround if there are no other VCHs running in vCenter Server that are configured with an operations user, as this removes the roles for all VCHs that are running in that vCenter Server instance.

  • Containers time out while starting with the error context deadline exceeded. #4294
    When this occurs the container VM is not powered off but is left in the Starting state, and might not have a configured network interface. Also, docker-compose and other tools that perform operations based on container state might not handle starting correctly. In the case of docker-compose it does not stop the container before trying to remove it.

  • Container start failure handled inconsistently. #4294

  • Deleting multiple VCHs fails. #7020
    Deleting multiple VCHs at the same time fails with one of the following errors:

    • vic/pkg/trace.(*Operation).Err: vic-machine-linux error: context deadline exceeded.
    • ServerFaultCode: The method is disabled by 'VIC'

    Workaround: Try the delete operation again.

  • vic-machine debug --rootpw option enables SSH. #6402
    If you specify the vic-machine debug --rootpw option without also specifying --enable-ssh, SSH access to the VCH is still enabled. This behavior is incorrect and will be fixed in a future release.

  • Gateway information is missing from docker inspect. #6010
    If you configured a container network on a VCH, if this network uses DHCP, and if you did not specify --container-network-gateway, the gateway is configured correctly but does not show up in the output of docker inspect commands that are run on containers on that network.

  • Schema 2 image manifests not supported. #5187
    vSphere Integrated Containers Engine does not support pulling images that only have schema 2 image manifests. If an image registry offers only the schema 2 manifest for an image, pull operations fail. This is most commonly seen when pulling an image by digest.

  • Container VMs deleted between upgrading and rolling back an upgrade on a VCH show up in docker ps. #5754
    If you upgrade a VCH, perform vic-machine delete on container VMs, then use vic-machine upgrade --rollback to revert the VCH to the previous version, the cache information in the VCH is reverted. However, the deleted container VM cannot be reverted, so if you run docker ps, the deleted container VMs are listed.

    Workaround: Restart the VCH endpoint VM in the vSphere Client.

  • docker diff does not fully work with all containers. #6059
    Running docker diff on certain containers, for example postgres, results in Error response from daemon: Server error from portlayer: Server error from portlayer: Server error from archive reader for device.

  • docker info does not report secure registries. #6256
    Running docker info on a VCH reports any registries that you have whitelisted, but does not report secure registries that you specified with vic-machine create --registry-ca.

  • Enabling content trust in Management Portal does not work if a VCH is in whitelist mode. #6258
    If you install a VCH with whiltelist mode enabled, attempting to a enable content trust on a vSphere Integrated Containers Registry in Management Portal does not update the whitelist, and you cannot pull from that registry.

  • Publishing all exposed ports to random ports with the -P option is not supported. #3000
    vSphere Integrated Containers Engine does not support docker create/run -P.

  • Occasional disconnection during vMotion. #4484
    If you are attached to a container VM that is migrated by vMotion, the SSH connection to the container VM might drop when vMotion completes.

    Workaround: Perform docker attach after the vMotion completes to reattach to the container.

  • Using volume labels with docker-compose causes a plugin error. #4540
    Setting a label in a volume in the Docker compose YML file results in error looking up volume plugin : plugin not found.

    Workaround: Set the volume driver explicitly as local or vsphere in the compose file. E.g.,

    volumes:
  volume_with_label:
    driver: local

  • vSphere Integrated Containers Management Portal cannot pull images from an insecure vSphere Integrated Containers instance when creating a container using vSphere Integrated Containers Engine. #4706
    Creating a container in vSphere Integrated Containers Management Portal with vSphere Integrated Containers Engine as the only Docker host results in the error certificate signed by unknown authority.

    Workarounds: Specify the vSphere Integrated Containers Registry port when you set the vic-machine create--insecure-registry option, or provide a CA certificate in the --registry-ca option.

  • Image store is in the wrong directory if the datastore already has a directory with the same name. #3365
    If the datastore already has a directory with the same name as the VCH, and the directory does not have a VM, vic-machine creates the VCH correctly names the folder a slightly different name. Example, folder "test_1" with vch named "test". The kvstore is located in "test_1" folder correctly, but image files are still in the "test" directory.

  • Deployment with static IP takes a long time. #3436
    If you deploy a VCH with a static IP, the deployment might take longer than expected, resulting in timeouts.
    Workaround: Increase the timeout for the deployment when using static IP.

  • Firewall status delayed on vCenter Server. #3139
    If you update the firewall rules on an ESXi host to allow access from specific IP addresses, and if that host is managed by vCenter Server, there might be a delay before vCenter Server takes the updated firewall rule into account. In this case, vCenter Server continues to use the old configuration for an indeterminate amount of time after you have made the update. vic-machine create can successfully deploy a VCH with an address that you have blocked, or else fail when you deploy a VCH with an address that you have permitted.

    Workaround: Wait a few minutes and run vic-machine create again.

  • When you pull a large sized image from Harbor into a virtual container host, you get an error that the /tmp partition reached capacity. #3624

    docker: Failed to fetch image blob: weblogic/test_domain/sha256:3bf21a5a3fdf6586732efc8c64581ae1b4c75e342b210c1b6f799a64bffd7924 returned download failed: write /tmp/3bf21a5a3fdf346188145: no space left on device.

    Workaround: Deploy the virtual container host with --endpoint-memory=4096 which increases the appliance memory configuration.
  • Pulling all tagged images in a repository is not supported. #2724

    vSphere Integrated Containers only attempts to pull the latest tagged images.
  • Container VM fails to start on VIC backed by a VVOL datastore. #2242

    VVOL datastores are not supported in this release.
  • Attaching the same container from multiple terminals causes problems. #2214
  • --net=none is not supported. #2108
  • VCH restarts if required process cannot be restarted. #2099

    The system attempts to restart a finite number of times, then reports an error, leaving the VCH up and running to download logs. Instead, VCH immediately reboots.
  • vic-machine incorrectly assumes conf.ImageStores[0] is the appliance datastore. #1884
  • When some of the hosts in the cluster are not attached to the dVS and do not have access to the bridge network, the error message is not easily readable. #1647
  • Image manifest validation for pulled images is not supported. #1331
  • Setting up overlay networks is not supported. #1222

    Error response from daemon: scope type not supported
  • vic-machine can connect to the target but the VCH appliance cannot. #3479

    The VCH cannot get an IP address on the management network or does not have a route to the specified target.
  • Adding folder options to vic-machine is not yet implemented. #773
  • Adding mapped vSphere networks to running containers is not yet implemented. #745
  • Adding bridge networks to running containers is not yet implemented. #743
  • Mapping an existing vSphere level network into the Docker network to explicitly provide a container with a route not through the VCH appliance is not yet implemented. #441
  • docker pull results an "already exists" error #1409

    If a context deadline exceeded error occurs on the port layer while performing an image pull, it causes an inconsistent state for the image. Pulls can also take a very long time with a slow network connection.
  • vic-machine create validation fails if a dvSwitch exists on an ESXi target #729

See also the known issues for each of the vSphere Integrated Containers components:

Open Source Components

The copyright statements and licenses applicable to the open source software components distributed in vSphere Integrated Containers Engine are available in the LICENSE file.

Assets 2
Loading