Hi,
I'm facing an issue with randomly disappearing logs on cloudwatch.
I have scheduled tasks on ECS which run container with defined inside entrypoint which is responsible for running task using this same container but with other entrypoint.
During first task(first entrypoint) - logging works correctly but when this task trigger another tasks(same container but other entrypoint) logs in cloudwatch appears randomly.
This happens even when I run that tasks by hand.
First run may fail with logs but the second one 5 minutes later works. With exactly the same settings.
I'm 100% sure that functionality inside container works(I checked).
I have no idea what kind of further investigation may I do.
Any suggestions?
EDIT:
I implemented boto3 for logs next to python logger.
Events via boto3 are handled correctly so it's a kind of workaround but it still not resolve an issue.
SOLUTION-WORKAROUND:
Use https://github.com/kislyuk/watchtower handler for python loggers
Edited by: karoljago on Apr 30, 2019 12:05 AM