Skip to content


This document lists several common problems. If you are looking for CPI specific errors see:

Getting logs from an unresponsive VM

If the BOSH agent on a VM becomes unresponsive, the bosh logs command will not work. This can make it difficult to diagnose what is causing the agent to become unresponsive.

To retrieve logs from an unresponsive VM:

  1. Launch a new VM within your IaaS. Ensure that you have remote access to the VM, such as through SSH.
  2. Run the bosh vms command to find the VM identifier for the unresponsive VM.
  3. Within your IaaS console, locate the unresponsive VM with the corresponding identifier, then:
    1. Shutdown the unresponsive VM.
    2. Detach the ephemeral disk from the unresponsive VM.
    3. Attach the ephemeral disk to the new VM.
    4. Reboot the new VM.
  4. SSH onto the new VM. Once on the VM:
    1. Switch to the root user.
    2. Use fdisk -l to identify the newly attached ephemeral disk from the unresponsive VM.
    3. Create a mount directory and use mount to mount the unresponsive VM's disk.
  5. You should now have access to the logs from the unresponsive VM. Logs for individual jobs can be located under <MOUNT-POINT>/var/vcap/sys/log/, while BOSH agent logs can be located under <MOUNT-POINT>/var/vcap/bosh/log.
  6. Once done examining or retrieving the relevant log files, you may delete the VM via the IaaS console. You may also need to delete the disks associated with the VM, depending on your IaaS provider.

For instructions on how to launch a new VM, shutdown or reboot VMs, attach or detach disks, and delete a VM, please refer to the documentation provided by the specific IaaS.

Timed out pinging to ... after 600 seconds

bosh deploy
  Failed creating bound missing vms > cloud_controller_worker/0: Timed out pinging to 013ce5c9-e7fc-4f1d-ac24 after 600 seconds (00:16:03)
  Failed creating bound missing vms > uaa/0: Timed out pinging to b029652d-14c3-4d68-98c7 after 600 seconds (00:16:12)
  Failed creating bound missing vms > uaa/0: Timed out pinging to 1f56ddd1-7f2d-4afc-ae43 after 600 seconds (00:16:23)
  Failed creating bound missing vms > loggregator_trafficcontroller/0: Timed out pinging to 28790bac-99a2-4703-89ad after 600 seconds (00:16:25)
  Failed creating bound missing vms > health_manager/0: Timed out pinging to 720b805b-928c-4bb7-b6dd after 600 seconds (00:16:52)
  Failed creating bound missing vms (00:16:53)

Error 450002: Timed out pinging to 013ce5c9-e7fc-4f1d-ac24 after 600 seconds

Task 45 error

For a more detailed error report, run: bosh task 45 --debug

This problem can occur due to:

  • blocked network connectivity between the Agent on a new VM and NATS (typically the Director VM). For stemcells newer than Xenial 621.178, Xenial 456.209, and Bionic 1.40, additional restrictions can block connectivity to NATS, specifically, traffic to the NATS endpoint is restricted to membership in a specific cgroup. To work around this restriction, follow these steps:
    # To add your SSH Session to the NATS Access CGROUP you can execute the below steps.
    # This will work EXCLUSIVELY for the SSH Session that executes the command and does
    # not persist across SSH Sessions/Connections.
    # Once SSH'ed onto a failing VM
    nc 4222 -v
    # (times out)
    sudo su
    source /var/vcap/bosh/etc/
    nc 4222 -v
    # Connection to 4222 port [tcp/*] succeeded!
    # INFO {....}
  • bootstrapping problem on the VM and/or wrong configuration of the Agent
  • blocked or slow boot times of the VM

It's recommended to start a deploy again and SSH into one of the VMs and look at the Agent logs while the Director waits for VMs to become accessible. See director.debug.keep_unreachable_vms property to let Director know to leave unreachable VMs for easier debugging.

Timed out sending 'get_task' to instance: 'unknown' ...

$ bosh deploy

Updating instance clock_global: clock_global/b0d76aa7-9a4e-44f6-8e28-72d941dc0e16 (0) (canary) (00:06:15)
  L Error: Timed out sending 'get_task' to instance: 'unknown', agent-id: 'f66bf7ce-bd5e-4528-937a-a25ba8223508' after 45 seconds
Error: Timed out sending 'get_task' to instance: 'unknown', agent-id: 'f66bf7ce-bd5e-4528-937a-a25ba8223508' after 45 seconds

In case when the VM spec has not been applied, the instance name is not available yet and the timeout error message will display that the instance name as 'unknown'. This also holds for other actions beyond get_task, such as get_state, cancel_task, apply, etc...

The steps for remediation are similar to the Timed out pinging to ... after 600 seconds case, where operators should try to SSH into VMs as the problem is occurring so they can look at the Agent logs. not running after update

bosh deploy
  Started updating job access_z1 > access_z1/0 (canary)
     Done updating job route_emitter_z1 > route_emitter_z1/0 (canary) (00:00:13)
     Done updating job cc_bridge_z1 > cc_bridge_z1/0 (canary) (00:00:20)
     Done updating job cell_z1 > cell_z1/0 (canary) (00:00:40)
   Failed updating job access_z1 > access_z1/0 (canary): `access_z1/0' is not running after update (00:02:13)

Error 400007: `access_z1/0' is not running after update

Task 47 error

For a more detailed error report, run: bosh task 47 --debug

This problem occurs when one of the release jobs on a VM did not successfully start in a given amount of time. You can use bosh instances --ps command to find out which process on the VM is failing. You can also access logs to view additional information.

This problem may also arise when deployment manifest specifies too small of a canary/update watch time which may not be large enough for a process to successfully start.

umount: /var/vcap/store: device is busy

L Error: Action Failed get_task: Task 5be893c6-7a2c-4f3f-420b-433fd23528a1 result: Migrating persistent disk: Remounting persistent disk as readonly: Unmounting /var/vcap/store: Running command: 'umount /var/vcap/store', stdout: '', stderr: 'umount: /var/vcap/store: device is busy.
        (In some cases useful info about processes that use
         the device is found by lsof(8) or fuser(1))
': exit status 1

This process occurs when one of the processes (from one of the installed jobs) did not fully shutdown and continues to retain a reference to the persistent disk. Agent tries to unmount persistent disk and unmount command fails. You can use bosh ssh command to get inside the VM and diagnose which process is holding onto the persistent disk via lsof | grep /var/vcap/store (ran as root).

Running command: bosh-blobstore-dav -c ... 500 Internal Server Error

bosh deploy

Failed compiling packages > dea_next/3e95ef8425be45468e044c05cc9aa65494281ab5: Action Failed get_task: Task bd35f7c1-2144-4045-763e-40beeafc9fa3 result: Compiling package dea_next: Uploading compiled package: Creating blob in inner blobstore: Making put command: Shelling out to bosh-blobstore-dav cli: Running command: 'bosh-blobstore-dav -c /var/vcap/bosh/etc/blobstore-dav.json put /var/vcap/data/tmp/bosh-platform-disk-TarballCompressor-CompressFilesInDir949066221 cd91a1c5-a034-4c69-4608-6b18cc3fcb2b', stdout: 'Error running app - Putting dav blob cd91a1c5-a034-4c69-4608-6b18cc3fcb2b: Wrong response code: 500; body: <html>
<head><title>500 Internal Server Error</title></head>
<body bgcolor="white">
<center><h1>500 Internal Server Error</h1></center>
', stderr: '': exit status 1 (00:03:16)

This problem can occur if the Director is configured to use built-in blobstore and does not have enough space on its persistent disk. You can redeploy the Director with a larger persistent disk. Alternatively you can remove unused releases by running bosh clean-up command.

If bosh clean-up command fails with 500 Internal Server Error, consider removing /var/vcap/store/director/tasks to free up a little bit of persistent disk space before running bosh clean-up command again. (Deleting that directory will delete Director task debug logs.)

Debugging Director database

It may be necessary to dive into the Director DB. The easiest way to do so is to SSH into the Director VM and use console. For example:

ssh vcap@DIRECTOR-IP


Should result in an array listing the current BOSH deployments, e.g.:

=> ["cf-dev", "cf-staging", "cf-prod"]


It's not recommended to modify the Director database via this or other manual methods. Please let us know via GitHub issue if you need a certain feature in the BOSH CLI to do some operation.

Task X cancelled

bosh deploy

  Started preparing package compilation > Finding packages to compile. Done (00:00:01)
  Started preparing dns > Binding DNS. Done (00:00:05)

Error 10001: Task 106 cancelled

Task 106 cancelled

This problem typically occurs if the Director's system time is out of sync, or if the Director machine is underpowered.

Upload release fails

bosh upload release blah.tgz
  Started creating new packages > blah_package/f9098f452f46fb072a6000b772166f349ffe27da. Failed: Could not create object, 413/<html>
<head><title>413 Request Entity Too Large</title></head>
<body bgcolor="white">
<center><h1>413 Request Entity Too Large</h1></center>

Error 100: Could not create object, 413/<html>
<head><title>413 Request Entity Too Large</title></head>
<body bgcolor="white">
<center><h1>413 Request Entity Too Large</h1></center>

This failure occurs due to nginx configuration problems with the director and the nginx blobstore. This can be remedied by configuring the max_upload_size property on the director and blobstore jobs.

Persistent Disk with id not found

bosh create-env
Command 'deploy' failed:
    Creating instance 'bosh/0':
      Updating instance disks:
        Updating disks:
          Deploying disk:
            Mounting disk:
              Sending 'get_task' to the agent:
                Agent responded with error: Action Failed get_task: Task 7e4d289d-b97c-4464-40d4-ecc90cc2a94b result: Persistent disk with volume id '14128b61-e046-48ae-b48a-fc0324716b83' could not be found

The SSH tunnel between your machine and the VM in the cloud can be terminated prematurly, see corresponding bug. Update CLI v2 to version >= v2.0.2 to fix this.

Errors creating or fetching credhub variables

Upgrading to Credhub 2.0.0 introduces a breaking change with permissions. Users no longer have automatic implicit read/write access for new paths. Any new credentials the director may generate as part of a deployment will not be created, and any variables the user did not previously have access to will be inaccessible.

See Credhub 2.0.0 release notes for more information.

Failed to find variable '/bosh/deployment/my_variable' from config server: HTTP Code '404', Error: 'The request could not be completed because the credential does not exist or you do not have sufficient authorization.'

Using the Credhub CLI, ensure reads & writes to existing paths, or new paths, work with current credentials.

Login with the $UAA_CLIENT and $UAA_CLIENT_SECRET where credhub.write is listed within the authorities. Clients for UAA are defined under the UAA job: properties.uaa.clients in the bosh director manifest.

credhub login --client $UAA_CLIENT --client-secret $UAA_CLIENT_SECRET

# When the user does not have adequate permissions
credhub set -n "/bosh/deployment/new_variable" -t password -w "test"
The request could not be completed because the credential does not exist or you do not have sufficient authorization.

# When the user has adequate permissions for this path
credhub set -n "/bosh/deployment/new_variable" -t password -w "test"

id: 1eccfea2-dad0-4a11-b8ab-68eae01de3ad
name: /bosh/deployment/new_variable
type: password
value: test
version_created_at: "X"

In order to create new credentials with Credhub 2.0.0 with ACLs enabled, operators must update their director manifests to include permissions. Below is typical configuration for an admin for all credential paths:

       - path: /*
         actors: ["uaa-client:credhub_admin_client"]   # this client must exist
         operations: [read,write,delete,read_acl,write_acl]

Otherwise, users can disable ACLs. Credhub recommends that ACLs are enabled and permissions are assigned explicitly.