github.com/pachyderm/pachyderm@v1.13.4/doc/docs/1.9.x/troubleshooting/general_troubleshooting.md (about)

     1  # General Troubleshooting
     2  
     3  Here are some common issues by symptom along with steps to resolve them. 
     4  
     5  - [Connecting to a Pachyderm cluster](#connecting-to-a-pachyderm-cluster)
     6    - [Cannot connect via `pachctl` - context deadline exceeded](#cannot-connect-via-pachctl-context-deadline-exceeded)
     7    - [Certificate error when using `kubectl`](#certificate-error-when-using-kubectl)
     8    - [Uploads and Downloads are slow](#uploads-and-downloads-are-slow)
     9  
    10  
    11  ---
    12  
    13  ## Connecting to a Pachyderm Cluster
    14  
    15  ### Cannot connect via `pachctl` - context deadline exceeded
    16  
    17  #### Symptom
    18  
    19  You may be using the pachd address config value or environment variable to specify how `pachctl` talks to your Pachyderm cluster, or you may be forwarding the pachyderm port.  In any event, you might see something similar to:
    20  
    21  ```
    22  $ pachctl version
    23  COMPONENT           VERSION                                          
    24  pachctl             1.9.5   
    25  context deadline exceeded
    26  ```
    27  
    28  Also, you might get this message if `pachd` is not running.
    29  
    30  #### Recourse
    31  
    32  It's possible that the connection is just taking a while. Occasionally this can happen if your cluster is far away (deployed in a region across the country). Check your internet connection.
    33  
    34  It's also possible that you haven't poked a hole in the firewall to access the node on this port. Usually to do that you adjust a security rule (in AWS parlance a security group). For example, on AWS, if you find your node in the web console and click on it, you should see a link to the associated security group. Inspect that group. There should be a way to "add a rule" to the group. You'll want to enable TCP access (ingress) on port 30650. You'll usually be asked which incoming IPs should be whitelisted. You can choose to use your own, or enable it for everyone (0.0.0.0/0).
    35  
    36  
    37  ### Certificate Error When Using Kubectl
    38  
    39  #### Symptom
    40  
    41  This can happen on any request using `kubectl` (e.g. `kubectl get all`). In particular you'll see:
    42  
    43  ```
    44  $ kubectl version
    45  Client Version: version.Info{Major:"1", Minor:"6", GitVersion:"v1.6.4", GitCommit:"d6f433224538d4f9ca2f7ae19b252e6fcb66a3ae", GitTreeState:"clean", BuildDate:"2017-05-19T20:41:24Z", GoVersion:"go1.8.1", Compiler:"gc", Platform:"darwin/amd64"}
    46  Unable to connect to the server: x509: certificate signed by unknown authority
    47  ```
    48  
    49  #### Recourse
    50  
    51  Check if you're on any sort of VPN or other egress proxy that would break SSL.  Also, there is a possibility that your credentials have expired. In the case where you're using GKE and gcloud, renew your credentials via:
    52  
    53  ```
    54  $ kubectl get all
    55  Unable to connect to the server: x509: certificate signed by unknown authority
    56  $ gcloud container clusters get-credentials my-cluster-name-dev
    57  Fetching cluster endpoint and auth data.
    58  kubeconfig entry generated for my-cluster-name-dev.
    59  $ kubectl config current-context
    60  gke_my-org_us-east1-b_my-cluster-name-dev
    61  ```
    62  
    63  ### Uploads and Downloads are Slow
    64  
    65  #### Symptom
    66  
    67  Any `pachctl put file` or `pachctl get file` commands are slow.
    68  
    69  #### Recourse
    70  
    71  If you do not explicitly set the pachd address config value, `pachctl` will default to using port forwarding, which throttles traffic to ~1MB/s. If you need to do large downloads/uploads you should consider using pachd address config value. You'll also want to make sure you've allowed ingress access through any firewalls to your k8s cluster.
    72  
    73  ### Naming a Repo with an Unsupported Symbol
    74  
    75  #### Symptom
    76  
    77  A Pachyderm repo was accidentally named starting with a dash (`-`) and the repository
    78  is treated as a command flag instead of a repository.
    79  
    80  #### Recourse
    81  
    82  Pachyderm supports standard `bash` utilities that you can
    83  use to resolve this and similar problems. For example, in this case,
    84  you can specify double dashes (`--`) to delete the repository. Double dashes
    85  signify the end of options and tell the shell to process the
    86  rest arguments as filenames and objects.
    87  
    88  For more information, see `man bash`.