Intermittent Failures With ARM64 Builds

[a1flecke]

We have noticed intermittent failures with installing Dynatrace OneAgent when using this buildpacks to build ARM images and running on ARM. Debug and verbose logs do not reveal anything helpful to us. Removing the pack caching directives or issuing the clear-cache directive does not clear it up.

Expected Behavior

Dynatrace OneAgent is installed

Current Behavior

If it matters: We are exploring migrating our existing AMD64 services to ARM64. The other buildpacks are fine. The Dynatrace buildpack is the only one that is experiencing this issue.

Pack build command:

pack build my-app:a-tag\
      --builder paketobuildpacks/builder-jammy-buildpackless-tiny\
      --buildpack paketo-buildpacks/amazon-corretto,paketo-buildpacks/java,paketo-buildpacks/dynatrace\
      --cache-image ***REDACTED**cache:pack-my-app\
      --clear-cache \
      --env ARTIFACTORY_USERNAME=$ARTIFACTORY_USERNAME \
      --env ARTIFACTORY_TOKEN=$ARTIFACTORY_TOKEN\
      --env BP_JVM_VERSION=21\
      --env BP_MAVEN_BUILD_ARGUMENTS="-Dmaven.test.skip=true --no-transfer-progress package --batch-mode"\
      --env BP_MAVEN_BUILT_ARTIFACT=backend/target/\
      --env BPE_DEFAULT_BPL_JVM_CLASS_ADJUSTMENT=120% \
      --env BPL_JMX_ENABLED=true \
      --verbose \
      --platform=linux/arm64\
      --env BP_LOG_LEVEL='DEBUG'\
      --volume $HOME/.m2:/home/cnb/.m2:rw\
      --volume $PWD/bindings/:/platform/bindings/:rw\
      --sbom-output-dir sbom \
      --publish

Sometimes it fails:

Running build for buildpack paketo-buildpacks/[email protected]
Looking up buildpack
Finding plan
Creating plan directory
Preparing paths
Running build command
Paketo Buildpack for Dynatrace 5.8.1
  https://github.com/paketo-buildpacks/dynatrace
  Dynatrace OneAgent 1.293.153.20240702-150912: Contributing to layer
  Warning: Dependency has no SHA256. Skipping cache.
    Downloading from https://**REDACTED***.live.dynatrace.com/api/v1/deployment/installer/agent/unix/paas/latest?bitness=64&skipMetadata=true&arch=arm&include=java
unable to invoke layer creator
unable to get dependency Dynatrace OneAgent. see DEBUG log level
Timer: Builder ran for 2m16.490621239s and ended at 2024-07-18T21:05:27Z
ERROR: failed to build: exit status 1
ERROR: failed to build: executing lifecycle: failed with status code: 51
real	2m57.405s
user	0m21.847s
sys	0m2.489s

Sometimes it passes:

Running build for buildpack paketo-buildpacks/[email protected]
Looking up buildpack
Finding plan
Creating plan directory
Preparing paths
Running build command
Paketo Buildpack for Dynatrace 5.8.1
  https://github.com/paketo-buildpacks/dynatrace
  Dynatrace OneAgent 1.293.153.20240702-150912: Contributing to layer
  Warning: Dependency has no SHA256. Skipping cache.
    Downloading from https://***REDACTED***.live.dynatrace.com/api/v1/deployment/installer/agent/unix/paas/latest?bitness=64&skipMetadata=true&arch=arm&include=java
    Expanding to /layers/paketo-buildpacks_dynatrace/dynatrace-oneagent
    Writing env.launch/BPI_DYNATRACE_BUILDPACK_ID.default
    Writing env.launch/BPI_DYNATRACE_BUILDPACK_VERSION.default
    Writing env.launch/DT_CUSTOM_PROP.append
    Writing env.launch/DT_CUSTOM_PROP.delim
    Writing env.launch/DT_LOGSTREAM.default
    Writing env.launch/LD_PRELOAD.delim
    Writing env.launch/LD_PRELOAD.prepend
  Launch Helper: Contributing to layer
    Creating /layers/paketo-buildpacks_dynatrace/helper/exec.d/properties
Processing layers
Updating environment
Reading output files
Updating buildpack processes
Updating process list
Finished running build for buildpack paketo-buildpacks/[email protected]

Possible Solution

Not sure

Steps to Reproduce

• The above pack build on a simple java service.

Motivations

• We want to migrate our hundreds of existing AMD64 ECS services to Gravitron (ARM64)

[dmikusa]

  Warning: Dependency has no SHA256. Skipping cache.
    Downloading from https://**REDACTED***.live.dynatrace.com/api/v1/deployment/installer/agent/unix/paas/latest?bitness=64&skipMetadata=true&arch=arm&include=java
unable to invoke layer creator
unable to get dependency Dynatrace OneAgent. see DEBUG log level

There is no pre-bundled archive we can use with Dynatrace. We always have to fetch the archive from Dynatrace and then install it. What's happening here is that the buildpack is attempting to fetch the dependency from Dynatrace and it's failing to download. It could be network-related, or it could be related to the server responding slowly.

When you pack build --env BP_LOG_LEVEL='DEBUG' -v ... does it give you any more details?

[a1flecke]

  Warning: Dependency has no SHA256. Skipping cache.
    Downloading from https://**REDACTED***.live.dynatrace.com/api/v1/deployment/installer/agent/unix/paas/latest?bitness=64&skipMetadata=true&arch=arm&include=java
unable to invoke layer creator
unable to get dependency Dynatrace OneAgent. see DEBUG log level

There is no pre-bundled archive we can use with Dynatrace. We always have to fetch the archive from Dynatrace and then install it. What's happening here is that the buildpack is attempting to fetch the dependency from Dynatrace and it's failing to download. It could be network-related, or it could be related to the server responding slowly.

When you pack build --env BP_LOG_LEVEL='DEBUG' -v ... does it give you any more details?

The above logs were with debug and verbose logging. There were no helpful details.

[dmikusa]

So, it's hard to say. It seems network or related to the dynatrace server response. You can try adjusting the timeout settings, and see if that might help. They correlate to various Go HTTP Client timeouts. I put some info below.

• BP_DIALER_TIMEOUT (default 6s)

  // Timeout is the maximum amount of time a dial will wait for
  // a connect to complete. If Deadline is also set, it may fail
  // earlier.

• BP_DIALER_KEEP_ALIVE (default 60s)

  // KeepAlive specifies the interval between keep-alive
  // probes for an active network connection.
  // If zero, keep-alive probes are sent with a default value
  // (currently 15 seconds), if supported by the protocol and operating
  // system.

• BP_TLS_HANDSHAKE_TIMEOUT (default 5s)

  // TLSHandshakeTimeout specifies the maximum amount of time to
  // wait for a TLS handshake. Zero means no timeout.

• BP_RESPONSE_HEADER_TIMEOUT (default 5s)

  // ResponseHeaderTimeout, if non-zero, specifies the amount of
  // time to wait for a server's response headers after fully
  // writing the request (including its body, if any). This
  // time does not include the time to read the response body.

• BP_EXPECT_CONTINUE_TIMEOUT (default 1s)

  // ExpectContinueTimeout, if non-zero, specifies the amount of
  // time to wait for a server's first response headers after fully
  // writing the request headers if the request has an
  // "Expect: 100-continue" header. Zero means no timeout and
  // causes the body to be sent immediately, without
  // waiting for the server to approve.
  // This time does not include the time to send the request header.

I could perhaps see BP_RESPONSE_HEADER_TIMEOUT being too short if the server is taking a while to respond. Maybe also BP_DIALER_TIMEOUT, if it's not getting a quick TCP connection (I believe DNS resolution is also included in this, so if DNS is slow that can throw off the connect timeout too).

Hope that helps!