-
Notifications
You must be signed in to change notification settings - Fork 11
Request cannot span more than 24 hours Error #61
Comments
Issue-Label Bot is automatically applying the label Links: app homepage, dashboard and code for this bot. |
Looking at the
|
Also a related error in other RoboMaker ROS node that is seems to be related to time if it helps. aws-robotics/cloudwatchmetrics-ros1#46 |
Hey @mr337 Thanks for filing this issue as well. Can you please verify that there is no stale data in the |
Following up from aws-robotics/cloudwatchmetrics-ros1#46 (comment) logs we like to think are more important and would like to always be uploaded. This looks like the way the logs are being pushed up is in a payload that has log events with timestamps greater than 24hr thus CWL rejects the payload. As a work around would reducing |
Given the documentation of the PutLogEvents, I would expect log data to be accepted if it's up to 2 weeks old. If the system was unable to upload logs for a period of time greater than two weeks then currently we are unable to upload :-( Since the data in How does this sound? |
TIL that CWL doesn't except logs older than 2 weeks as well. I think for our application we are OK with that. The concern I have is error on the the log node seems to be when uploading logs that are within the 2 week period. For example these files are within the last 4ish days
And still receive the same
All those logs are within the 2 week and should be able to be uploaded. |
You're absolutely right. Please let me know if you can't workaround this issue by setting a smaller batch upload size and the proposed options below. I think we can try a few other things with the config pararms:
|
I will try that. I'm wasn't for sure if I was interpreting those config params correctly. Will report back in a few days. |
I have been doing some testing to try to find a good set of parameters and so far no luck. Here is what I have tried: The sending of cached logs does not seem to honor the publish frequency and then runs into the CW Log API throttle.
Setting anything larger than 10 I have hit a 24 hour period issue as before which is odd. Here is a test with a logs file a week old.
The ROS log of it failing on 24hour when
My guess is the |
@mr337 that is unfortunate: I think I have an idea why and we will investigate further. Thanks for attempting the workaround and providing your results in great detail! |
@mr337 Sorry for the delay: we are actively looking at this issue. |
No problem, thanks for the update. |
Hi @mr337, We've recently pushed fixes for aws-robotics/cloudwatch-common#53 and aws-robotics/cloudwatch-common#54 to a dependency of this project. If you're installing from source then you should see the changes when #64 is merged. Otherwise we can keep you updated on when we bloom-release the fixes to the packages. |
Awesome and thank you! |
I am starting to see odd errors in our logs from CloudWatch Log node when running the application live.
What is really odd is after this error we still get application shipped to CloudWatch and have been using them for weeks.
Things I have checked:
Using
Version: 2.3.1-1xenial-20200330-153719+0000
The text was updated successfully, but these errors were encountered: