Can someone help me figure out what causes these `...
# ask-community
s
Can someone help me figure out what causes these
UnixHTTPConnectionPool
errors in Prefect Cloud? These happen seemingly at random. I have this particular flow set to retry 3 times and it I got this on all 3 retries.
k
Hey @Sean Leakehe, I have never seen this before. Is the Prefect Cloud or Server?
s
Hey Kevin, this is Prefect Cloud.
k
You’re using docker agent?
s
Yep
k
Are you using a volume mount? Are you on a Mac?
Have you seen this thread?
👀 1
or you can try bumping up the read timeout setting?
s
All of our flows run on an EC2 running CentOS.
Sorry, I’m a bit new to this, the read timeout is a docker setting ,correct?
k
Yeah
1
Like this
🙏 1
Seems like it could be a load thing? Are you starting up a lot of containers when you see this error?
s
I think there could be as many as 4 containers that get spun up at the same time.
k
I think it’s hard to guess the impact of 4 containers without knowing the resources of the machine, but the comments of the StackOverflow link I sent seem to suggest bumping up resources may help (in addition to the timeout settings)?
s
Roger that. I’ll start out with testing an increased timeout. Thanks for the help!
👍 1
Looping back - We were able to increase the timeout by setting
--docker-client-timeout INTEGER
when starting the docker agent.
k
Ah nice! Thanks for the info
👍 1