Messages passed visibility timeout still "in flight"

Question

I am occasionally seeing a problem that I cannot explain with Fifo queues.  
In the console and on the aws-cli I see messages in-flight that are way passed their visibility timeouts.  
They stay there for hours.  
It seems to be the same issue this person posted:  
https://forums.aws.amazon.com/thread.jspa?messageID=520092  
  
Looking at my process output, all messages seem to have been processed without issue so I don't know what the remaining in flight means.  
I am using boto3 with python2 to interact with the queues.  
  
Thank you

Accepted Answer

Hi there!  
  
What you're describing sounds more like there are a few messages that are repeatedly failing to process for some reason. If you don't have a redrive policy () defined for a FIFO queue, any messages that can't be processed will cycle back and forth between visible and invisible indefinitely.  
  
If you're looking at metrics like ApproximateNumberOfMessagesNotVisible, it could look like it's staying elevated consistently, giving the impression that messages are staying invisible for a long time, when in truth there are several poison pills fluctuating back and forth.  
  
What does ApproximateAgeOfOldestMessage look like for your queue?  
  
Robin

Messages passed visibility timeout still "in flight"

関連するコンテンツ