Issue using 0.11.1: "Transient error [...] while exporting traces"

[rgilton]

Running a simple script with 0.11.1 results in lots of error messages like this:

Transient error StatusCode.UNAVAILABLE encountered while exporting traces to localhost:4317, retrying in 1s.
Transient error StatusCode.UNAVAILABLE encountered while exporting traces to localhost:4317, retrying in 2s.
Transient error StatusCode.UNAVAILABLE encountered while exporting traces to localhost:4317, retrying in 4s.
Transient error StatusCode.UNAVAILABLE encountered while exporting traces to localhost:4317, retrying in 8s.
Transient error StatusCode.UNAVAILABLE encountered while exporting traces to localhost:4317, retrying in 16s.
Transient error StatusCode.UNAVAILABLE encountered while exporting traces to localhost:4317, retrying in 32s.

These error messages don't happen with 0.10.3. Anyone know if this is a bug, or have I misconfigured something?

[wingyplus]

Did you run it with dagger run or just executing the script (or with python) directly?

[rgilton]

With python directly.

[helderco]

Sorry about that @rgilton! It's a bug. I thought dagger run was setting the telemetry env vars but it doesn't. I've actually been working on that this week so it should be fixed by next release.

I'll see if there's an env var you can use to disable it in the meantime.

[helderco]

The script I am running is in the gist that I linked to in the original question.

Quick tip: you don't need to pass the client around instance anymore, see the Global Client.

I don't know if this is a feature of 0.11.1, or something to do with the logging...

It's from dagger 0.11, not Python specifically. You should see some progress while it's happening and in the end it goes away to leave only the script's own output.

If you run the script through dagger run you have more control. You can increase the verbosity with -v (up to -vvv):

dagger run -v python foo.py

Which will persist a small amount of that progress after the script finishes.

You can also return to the previous output by disabling the TUI:

dagger run --progress=plain python foo.py

[johnstonematt]

So, in 0.11, there is no way of getting the "complete" output without running dagger run?

If that is the case could I put in a huge request that that be reverted? The output from simply running python foo.py in 0.10 and 0.9 was incredibly helpful.

[rgilton]

Quick tip: you don't need to pass the client around instance anymore, see the Global Client.

😃

You should see some progress while it's happening and in the end it goes away to leave only the script's own output.

Yea, I saw literally that one line flash up and no other lines!

Ok, if I run it via dagger run in the way you say (didn't know I could do that!) then I do get more output.

[rgilton]

It was also useful to be able to run the script without having had to install the dagger binary separately: if it's just in a python script, it's just a pip install dagger-io to get it to work. Whereas if dagger has to be around then there has to be some way of installing that first...

[helderco]

If that is the case could I put in a huge request that that be reverted?

Best create an issue asking for it. There was a similar issue here for dagger run:

• 🐞 --progress=plain stopped streaming stdout with Dagger v0.11.0 #7057

The difference is you need it on an SDK's automatic provisioning (it's the SDK downloading and executing the CLI underneath).