Lambda Using VPC Calling 3rd Party API Intermittently Experiencing Timeout Since 22/10/2025

Question

Multiple customers with AWS Lambda functions are experiencing intermittent TCP connection failures (SYN timeout) when calling the same 3rd party API since midday on 22nd October. The 3rd party's firewall never sees the requests. This affects multiple AWS accounts and multiple customers of the 3rd party (including us). Other customers, coming from Azure are not seeing a problem.

I believe there is a problem routing between AWS’s Lambda egress network (Hyperplane) and the 3rd party's ASNs.

Can anyone suggest any further steps to help rectify this or alternatively, is this something that is out of my control? If so, who might be the go-to person or organisation to tackle this?

Shahed Kazi (@shajam) · Answer

I can tell you the steps necessary for troubleshooting.
- Are the Lambda functions using your VPC / subnets. Can you check if the security groups, NACLs and VPC routing has not changed. It's possible that the Lambda was running in one subnet before but now running in multiple and it's failing sometimes when running in the alternate subnet.
- Can you check the VPC flow logs and check for the egress traffic. See if the traffic appears and check the outcome.

Ben · Answer

Thanks for responding.

Nothing has changed in our environment. It's worth reiterating that this same problem is occurring for other customers that attempt to use the 3rd party's API. It also started to occur in our completely separate test environment too. However, it is only occurring for customers using AWS. Going direct from a laptop works.

I have begun recording VPC flow logs and will check those as API calls are made today.

Lambda Using VPC Calling 3rd Party API Intermittently Experiencing Timeout Since 22/10/2025

Add your answer

Relevant content