Pipeline steps bigger than 8GB?

[neilscales]

I'm using .net 8 AOT to publish my AWS Lambda, and whilst it publishes OK locally, when I put it onto the build server, I get 'Killed' with signal 137.  I understand from other posts that this is a memory issue.

I've already got an 8GB docker image.  Is there any way to increase this?

definitions:
services:
docker:
# Define the Docker service which requires 7128 MB memory
docker-7128m:
type: docker
memory: 7128


<snip>

- step:
script:
# Build Lambda Login image
- dotnet publish XXXX
   --framework "net8.0"
   --self-contained true
   /p:GenerateRuntimeConfigurationFile=true
   --runtime linux-x64
   /p:StripSymbols=true
services:
- docker-7128m

.NET AOT builds are renowned for requiring huge amount of RAM, so there's not much I can do about it, apart from turn if off, which is obviously not going to happen.

[Theodora Boudale · Accepted solution]

Hi @neilscales and welcome to the community!

The maximum memory that can be configured for a step running on Atlassian's infrastructure is 8 GB with size: 2x. Therefore, the maximum memory that can be allocated to a service is 7 GB.

However, we will soon provide the option for larger step sizes (4x and 8x, with 16 GB and 32 GB of memory respectively). We will release this feature in June:

• https://bitbucket.org/blog/evolving-bitbucket-pipelines-to-unlock-faster-performance-and-larger-builds

You can add yourself as a watcher to the following feature request (by selecting the Start watching this issue link) if you'd like to get notified via email when this is released:

• https://jira.atlassian.com/browse/BCLOUD-17260

In the meantime, if you have a server where you can run this build, you could use one of our self-hosted runners that already allow you to configure 4x or 8x size steps (and you can therefore allocate more memory to the docker service):

• https://support.atlassian.com/bitbucket-cloud/docs/runners/

Kind regards,
Theodora