Where communities thrive

  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
Repo info
  • Jan 25 2019 16:23

    ark3 on ssh-cleanup


  • Jan 25 2019 16:23

    ark3 on master

    Return status rather than crash… Emit a distinct crash message f… Factor out sshuttle command and 3 more (compare)

  • Jan 25 2019 16:23
    ark3 closed #912
  • Jan 25 2019 14:57
    ark3 opened #912
  • Jan 25 2019 14:56

    ark3 on ssh-cleanup

    Return status rather than crash… Emit a distinct crash message f… Factor out sshuttle command and 2 more (compare)

  • Jan 25 2019 12:24
    bartoszmajsak opened #911
  • Jan 24 2019 22:04
    ark3 labeled #910
  • Jan 24 2019 22:04
    ark3 opened #910
  • Jan 24 2019 21:42
    ark3 closed #807
  • Jan 24 2019 21:42

    ark3 on master

    Don't pass --v=4 to kubectl eve… (compare)

  • Jan 24 2019 21:30
    ark3 closed #889
  • Jan 24 2019 21:30

    ark3 on master

    fix: changes depracated method … fix: masks logged tokens chore: adds changelog file and 8 more (compare)

  • Jan 24 2019 21:30
    ark3 closed #890
  • Jan 24 2019 21:29
    ark3 synchronize #890
  • Jan 24 2019 21:25
    ark3 closed #812
  • Jan 24 2019 21:25
    ark3 commented #812
  • Jan 24 2019 21:23
    ark3 closed #735
  • Jan 24 2019 21:23
    ark3 commented #735
  • Jan 24 2019 20:52
    rohansingh commented #905
  • Jan 24 2019 20:50
    ark3 commented #905
<ark3> Would you mind running an experiment for me? Can you try git checkout v1.3 and then pick things up starting at docker build? That will roll back to an older version of QotM.
<tuckerwales> Sure 🙂
<ark3> Thanks, and sorry about the broken tutorial.
<tuckerwales> No problem at all...
<tuckerwales> Yup, that works @ark3. Was it Flask not being in debug mode?
<ark3> Exactly that.
<tuckerwales> Ah brilliant, we're just looking into Kubernetes/Telepresence so I didn't know if it was something else...
<tuckerwales> Thanks a lot for your help, much appreciated!
<ark3> Happy to help. I'll fix this tutorial for the 0.98 release, which will hopefully happen today.
<tuckerwales> Awesome - thanks for Ambassador/Telepresence too. We're looking to integrate both offerings, they both look very good.
<ark3> Excellent. The folks in #ambassador are very responsive. Let us know how it goes.
<tuckerwales> Will do 🙂
<crcornwell> hey all! i'm using telepresence to develop against a minikube cluster on an aws ec2, and i'm running into an issue opening headless chrome. i have a python flask application with an endpoint that will take and save a screenshot of a webpage by creating a subprocess that runs a node script which uses puppeteer to open headless chrome and take the screenshot. when i run this flask application using gunicorn outside of telepresence and curl into it, it manages to create the screenshot with no problem, but running it with telepresence causes it to fail with the error: TimeoutError: Timed out after 30000 ms while trying to connect to Chrome! has anyone experienced anything similar, or have any ideas about things i could try?
<crcornwell> this is using telepresence 0.97, node 9.2.1, puppeteer 1.11.0, python 2.7, flask 1.0.2 and gunicorn 19.9.0, and running on Ubuntu 18.04. if there's any other information that would be useful please let me know
<crcornwell> well, of course i would figure it out as soon as i post here; i just noticed the message about static binaries not being supported when using --method inject-tcp, and sure enough running a telepresence shell without that flag allows the node command to complete just fine
<ark3> 🎉
<crcornwell> unfortunately i've run into some issues trying to use the default vpn-tcp method, for some reason if i swap a service deployment using that, many of my API calls fail with the error Unable to open PDO connection [wrapped: SQLSTATE[HY000] [2002] Connection refused]
<crcornwell> if i run with --method inject-tcp i don't get those errors, however
<ark3> The key difference is that inject-tcp redirects all network traffic from the affected processes, whereas vpn-tcp redirects only traffic it thinks is for the cluster, but for all processes.
<ark3> You may need an --also-proxy to work around that. Or consider using the container method, which also redirects all network traffic and works with any container.
<crcornwell> by the container method do you mean using --docker-run instead of --run? or is it something else?
<ark3> Yes, exactly.
<crcornwell> ok, interesting. i'll look into that, thanks!
<ark3> @crcornwell Your usage of Telepresence is complex and interesting. Is your work open source? Can you talk about it?
<crcornwell> unfortunately it's not open source, but i can talk about it a bit
<crcornwell> we actually just started using it, about a year ago we moved our infrastructure to kubernetes using amazon's EKS. prior to that we were just running our services manually on several machines we were renting in a datacenter
<crcornwell> after we migrated our production services to k8s, we didn't update our development environments, and so there was an increasing amount of friction between how we needed to configure our application so that it would work in production, vs how it needed to work on our dev environments
<crcornwell> so over the last month or so i've been tasked with figuring out how we could bring our development servers more in line with how things are running on production
<crcornwell> i set up a minikube cluster and deployed our application to that, and then i found telepresence which i think is what actually makes this viable for development
<crcornwell> needing to constantly rebuild images and deploy them into the cluster to to test changes seems like a pretty slow pace, and likewise execing into running pods and developing there feels really bad
<crcornwell> so when i found telepresence, and how it enables you to develop the code on your host's filesystem while still seeing those changes immediately reflected in the cluster, i felt like i had found some kind of holy grail
<crcornwell> not to mention that i think it will prove to be really valuable in debugging our production system as well, now that we're aware of it
<rdl> @crcornwell this is really interesting; do you think you'd be open to doing a case study with the CNCF for your use case?
<crcornwell> absolutely! we're still in early days here, so i think it would be a better case study once the team here has had a chance to really get their hands dirty with this new workflow, but at that point i'd be happy to
Blake Miller
@ark3 sorry for the delayed response, I was out of town for a while
by "magic" I meant that the AWS DNS service by design gives two different address responses for the same name, depending on a special property of the client making the request
if that client is inside an AWS data center, then it resolves certain names to their internal IPs rather than the external IP it would resolve to from the outside world
those names include the autogenerated ec2 instance ones like ec2-*.compute-1.amazonaws.com
Hi, waking up the conversation a bit
Is there anyway to make telepresence work on windows? Inject-tcp is very partial solution, and does not allow working with an IDEA(ww are using intellij)
Alexander Fedulin

Hey everyone! I don't understand how to use kubectl with a remote AWS EKS cluster while telepresence is running. I receive

Unable to connect to the server: http: server gave HTTP response to HTTPS client

Any suggestions?

Chris Rutherford
Hey I'm having an issue on Mac that when telepresence is swapping the deployment, all other http traffic stops. Is that normal? Can I fix that?
Blake Miller
T: Failed to create Pod: T: error: SchemaError(io.k8s.api.core.v1.ObjectReference): invalid object doesn't have additional properties As of this morning, we're all unable to run telepresence 0.108
anybody else seeing that error message?
deployment or pod, the result is the same
ah, colleague has fixed this ^ by updating kubectl to 1.11 (this is odd as our cluster is 1.10 and we did not make any changes)
Samuli Silvius

Hi, I try to run telepresence 0.108 inside a container. Same problem as here:

I'm using Visul Studio Code devcontainer development environment and my bas eimage is
FROM mcr.microsoft.com/vscode/devcontainers/python:0-3.8
(see https://github.com/microsoft/vscode-dev-containers/tree/v0.154.0/containers/python-3/.devcontainer/base.Dockerfile)

I have worked out already e.g. this /tmp mounting problem telepresenceio/telepresence#1196

And telepresence is able to start and swap the pod in remote cluster.
But when it tries to setup proxy network container locally and tries to ssh to remote cluster through it. It fails.

Looks like there's a bug in our code. Sorry about that!

Traceback (most recent call last):
  File "/usr/bin/telepresence/telepresence/cli.py", line 135, in crash_reporting
  File "/usr/bin/telepresence/telepresence/main.py", line 81, in main
    user_process = launch(
  File "/usr/bin/telepresence/telepresence/outbound/setup.py", line 157, in launch
    return run_docker_command(
  File "/usr/bin/telepresence/telepresence/outbound/container.py", line 174, in run_docker_command
    raise RuntimeError("SSH to the network container failed to start.")
RuntimeError: SSH to the network container failed to start.

Here are the last few lines of the logfile (see /workspaces/neo-e2e/scripts/telepresence.log for the complete logs):

  39.2  22 |   29.8 TEL | [118] Running: ssh -F /dev/null -oStrictHostKeyChecking=no -oUserKnownHostsFile=/dev/null -oConnectTimeout=5 -q -p 38023 telepresence@ /bin/true
  39.2  22 |   29.8 TEL | [118] exit 255 in 0.00 secs.
  39.2  22 | Traceback (most recent call last):
  39.2  22 |   File "/usr/bin/entrypoint.py", line 130, in <module>
  39.2  22 |     main()
  39.2  22 |   File "/usr/bin/entrypoint.py", line 66, in main
  39.2  22 |     proxy(loads(sys.argv[2]))
  39.2  22 |   File "/usr/bin/entrypoint.py", line 86, in proxy
  39.2  22 |     "SSH from local container to the cluster failed to start."
  39.2  22 | RuntimeError: SSH from local container to the cluster failed to start.
  39.2  22 | [INFO  tini (1)] Main child exited normally (with status '1')
  39.4 TEL | [22] Network container: exit 1

Would you like to file an issue in our issue tracker? You'll be able to review and edit before anything is posted to the public. We'd really appreciate the help improving our product. [Y/n]:

I running this on Ubuntu 20.04.
in devcontainer I have mounted docker socket "source=/var/run/docker.sock,target=/var/run/docker.sock,type=bind", So using host docker engine.
and running devcontainer with "--privileged" docker run option.
Any idea if this setup could work?

Jakob Hostnik
has anyone had this problem on arch linux?
I tried telepresence 2.1.1. for linux and when I try to connect to minikube. I get only
Launching Telepresence Daemon v2.1.1 (api v3)
Need root privileges to run "/usr/local/bin/telepresence daemon-foreground /home/jakob/.cache/telepresence/logs '' ''"
[sudo] password for jakob:
Connecting to traffic manager...
Telepresence Daemon quitting...done
telepresence: error: context canceled
Hey all!
I tried to install telepresence on my Mac (Catalina).
I have done brew install --cask osxfuse since the command in the doc is not working for me.
this is installing osxfuse, then I did brew install datawire/blackbird/telepresence and I get the following error:
==> Installing telepresence from datawire/blackbird libfuse: Linux is required for this software. Error: An unsatisfied requirement failed this build.
Can you please help make it works?
I installed version 2 in the end, will try to work with it