Hi everyone, I'm trying to use airflow 2.4.3 with ...
# troubleshoot
b
Hi everyone, I'm trying to use airflow 2.4.3 with meltano's datahub plugin, but everytime I try to start up airflow I'm now getting an error that says:
Copy code
Invalid version: '[<current_datetime>] {_plugin.py:349} INFO - Patching datahub policy 2.4.3'
I'm trying to use
apache-airflow==2.4.3
,
apache-airflow-providers-amazon==6.0.0
,
acryl-datahub-airflow-plugin==0.9.6.2
. I've configured the
gms_host
in the datahub utility, but I haven't set the connection on the airflow side because I would normally do this through the UI. Is there a way for me to set the configuration in my
meltano.yml
? I've tried setting
datahub_conn_id
under
config.core
but no luck. Any idea what policy this error is referring to?
b
hey Jam! i'm asking around to find the right person to answer your question, we'll get back to you before too long 🙂 appreciate your patience
b
Thank you so much, Chris!
It looks like the error is happening here:
It doesn't happen when I don't have
acryl-datahub-airflow-plugin
installed
b
okay good to know! also our airflow expert isn't around right now but said he will help you out first thing tomorrow morning if that's alright?
b
That works for me. Thank you!
d
@bulky-jackal-3422 hey, Unfortunately, I don’t know much about Meltano, but the config should be the Airflow config, where you can set those configuration properties. I don’t know how you can change it when you use Meltano? Do you have direct access to Airflow, or Meltano has an embedded Airflow? How is this working? What is the stack trace in the above screenshot? Is it fail there? As the stderr is empty there and the stdout shows a log message which is expected (it is the message about it trying to patch all the airflow dag/tasks to add success/failure hooks automatically)
b
Hi @dazzling-judge-80093! 1. Meltano passes the config I put in the
meltano.yml
to the
airflow.cfg
. 2. I'm invoking Airflow through Meltano. I can find out how to access Airflow directly. When I normally run Airflow via Meltano and I do
docker ps
, I don't see any containers running, so I'll have to find another way. 3. The stacktrace is via Meltano's debug log. I screenshotted that specific part, because it was the only place I could find the mention of
version
and since the stacktrace I posted initially mentions
Invalid version
it piqued my interest. Is there any interaction between
acryl-datahub-airflow-plugin
and the version of Airflow?
d
ahh, I think I know what is the problem
As you can see somehow, somewhy meltano try to get a version based on what is on the stdout. Our plugin prints out a
info
level log about it is patching the Airflow global policy -> https://github.com/datahub-project/datahub/blob/31260888fc0ce5135d3fe3bb89decae63926d418/metadata-ingestion/src/datahub_provider/_plugin.py#L349 but because of this Meltano picks this string up and try to parse as a version (it is a bit strange to me why they do it like this) and it fails to parse this string as a version.
b
Okay, that's what I thought was going on too, but I haven't heard back from the Meltano folks to see why it's handled this way
d
If you can change the logger, then maybe you can silence the DataHub plugin or in the long term, we can as well change this logline to debug level if it helps or provides a way to silence it
b
I'll try changing the log level to see if I can move past this issue!
When I changed that line that you mentioned to debug, it all worked
d
ahh, cool
b
Changing my log-level didn't seem to make an impact, though
d
ok, I think we can change as well this to debug to make sure it won’t break Meltano
b
Awesome, I really appreciate the help!