bitbucket pipeline produces a broken docker image

[AZZ]

Hello,

the docker image created by a pipeline crashes during execution. the image created on a local box - runs fine.

Here are the details:

bitbucket-pipelines.yml

image: atlassian/default-image:3
definitions:
services:
docker-2048:
memory: 2048
type: docker
steps:
- step: &Clean_cache
name: Clean Cache
script:
- pipe: atlassian/bitbucket-clear-cache:3.2.0
variables:
BITBUCKET_USERNAME: $BITBUCKET_USERNAME
BITBUCKET_APP_PASSWORD: $BITBUCKET_USER_APP_TOKEN
CACHES: ["docker"]
 - step: &Build
name: Build
services:
- docker-2048
# caches:
# - docker
script:
- branch=$( echo "${BITBUCKET_BRANCH////_}" | tr '[:upper:]' '[:lower:]' )
- export TAG=$DOCKER_HUB_REPO:$branch_${BITBUCKET_BUILD_NUMBER}_${BITBUCKET_COMMIT}
- echo "export TAG=$TAG" > docker.vars
- export DOCKER_FILE="docker/Dockerfile.dev"
- echo "export DOCKER_FILE=${DOCKER_FILE}" >> docker.vars
- cat docker.vars
- docker build -t $TAG -f $DOCKER_FILE .
- docker images
- docker login -u $SERVICE_ACCOUNT -p $SERVICE_ACCOUNT_TOKEN
- docker push $TAG
pipelines:
custom:
Build:
- step: *Clean_cache
- step: *Build

the partial output is:

.....

Step 26/31 : COPY misc/config/*.cfg ${CONFIG_PATH}/

---> daa915364ce7

Step 27/31 : COPY misc/config/resource ${RESOURCE_PATH}

---> 4eeac5338d96

Step 28/31 : RUN sed -i '/\[jcli\]/a bind=0.0.0.0' ${CONFIG_PATH}/jasmin.cfg

---> Running in 6af8720ab465

Removing intermediate container 6af8720ab465

---> 9edde9f21b48

Step 29/31 : COPY docker/docker-entrypoint.sh /

---> 7eaab101038c

Step 30/31 : ENTRYPOINT ["/docker-entrypoint.sh"]

---> Running in 7d3dffe7b240

Removing intermediate container 7d3dffe7b240

---> 09dc3c59ff08

Step 31/31 : CMD ["jasmind.py", "--enable-interceptor-client", "--enable-dlr-thrower", "--enable-dlr-lookup", "-u", "jcliadmin", "-p", "jclipwd"]

---> Running in f57dfab06da0

Removing intermediate container f57dfab06da0

---> e0d7ea1446a5

Successfully built e0d7ea1446a5

Successfully tagged xxxxxx/xxxxxx:75_1f156f460865d4966e865d876e43e9edd92a1064

if i run the build locally:

`docker build -t local -f ./docker/Dockerfile.dev .`

then the output is:

...

=> [19/22] COPY misc/config/*.cfg /etc/jasmin/ 0.1s
=> [20/22] COPY misc/config/resource /etc/jasmin/resource 0.1s
=> [21/22] RUN sed -i '/\[jcli\]/a bind=0.0.0.0' /etc/jasmin/jasmin.cfg 0.5s
=> [22/22] COPY docker/docker-entrypoint.sh / 0.1s
=> exporting to image 6.2s
=> => exporting layers 6.2s
=> => writing image sha256:c4f98aa31ecd3c145d9b5fd8a9ffa0961046dc1bc8979a8799739deb2db4c512 0.0s
=> => naming to docker.io/library/local

aside from the fact that the output style is very different ( i give it to the difference between build and buildkit ), the number of steps in the build is different. At the same the steps seemed to be identical in nature.

The sizes of the images look identical:

[az@dell5000 SMSMS-48]$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
local latest c4f98aa31ecd 2 hours ago 1.47GB
xxxxx/xxxx  76_c75e7ed055957032fd0dc8ed2987978336e1f97f c8e0d0e8ff90 3 hours ago 1.47GB

the build time is drastically different:

local build time is ~40 minutes, whereas BB gets job done in 6 minutes. I can theoretically attribute this different to the download time for all dependencies, but...

 I'm very confused about the situation. Appreciate your insight.

[Theodora Boudale · Accepted solution]

Hi @AZZ,

Could you please share some additional info so we can better understand the issue you mention below?

the docker image created by a pipeline crashes during execution

1. Are you starting a Docker container with that image locally?
2. What is the exact command you are using to start the container?
3. Does the container not start at all? Or does it start and it's crashing later?
4. Are you executing anything on this container before it crashes?
5. Does the command docker logs <container_id>, where <container_id> is the id of the container that crashes, show any output and errors?

Kind regards,
Theodora

[AZZ]

@Theodora Boudale,

 thank you for looking into this issue.

1. Are you starting a Docker container with that image locally?
   yes.
2. What is the exact command you are using to start the container?
   I use ` docker compose up jasmin`
   here is the docker-compose.yml:
   

   version: "3"
   
   services:
   redis:
   image: redis:alpine
   
   rabbit-mq:
   image: rabbitmq:alpine
   
   jasmin:
   build:
   context: ./
   dockerfile: ./docker/Dockerfile.dev
   image: xxxxxx/xxxxx:76_c75e7ed055957032fd0dc8ed2987978336e1f97f
   volumes:
   - ./jasmin:/usr/jasmin/jasmin
   ports:
   - 2775:2775
   - 8990:8990
   - 1401:1401
   depends_on:
   - redis
   - rabbit-mq
   environment:
   REDIS_CLIENT_HOST: redis
   AMQP_BROKER_HOST: rabbit-mq

   i use this same compose file for invoking both locally compiled and "bitbucket" compiled images, by replacing the image in the "image:" tag.
3. Does the container not start at all? Or does it start and it's crashing later?
   it starts, but then crashes with python specific error, when i access a certain api:

   smsms-48-jasmin-1 | File "/usr/local/lib/python3.11/site-packages/jasmin/protocols/cli/protocol.py", line 150, in lineReceived
   
   smsms-48-jasmin-1 | return self.sessionLineCallback(cmd, arg, line)
   
   smsms-48-jasmin-1 | File "/usr/local/lib/python3.11/site-packages/jasmin/protocols/cli/managers.py", line 14, in filter_cmd_and_call
   
   smsms-48-jasmin-1 | return fCallback(self, *args, **kwargs)
   
   smsms-48-jasmin-1 | File "/usr/local/lib/python3.11/site-packages/jasmin/protocols/cli/mtrouterm.py", line 103, in parse_args_and_call_with_instance
   
   smsms-48-jasmin-1 | RouteClassArgs = inspect.getargspec(self.sessBuffer['route_class'].__init__).args
   
   smsms-48-jasmin-1 | builtins.AttributeError: module 'inspect' has no attribute 'getargspec'
   
   smsms-48-jasmin-1 | 

4. Are you executing anything on this container before it crashes?
   I make API calls to the service provided by this container. Some work, but a specific one always seemed to crash the "bitbucket" built container.
5. Does the command docker logs <container_id>, where container_id is the id of the container that crashes, show any output and errors?
   please see above.
   
   I can provide (privately) all config files if you think this will help to troubleshoot more efficient.
   thank you!

[Theodora Boudale]

Hi @AZZ,

Thank you for the info, I believe it's best to create a support ticket so you can share more details privately.

I will go ahead and create it for you using the email of your community account and I'll include all the details you have shared here. You will receive shortly an email with a link to the support ticket, the ticket will be visible only to you and Atlassian staff. After you receive the email with the link, please post in the ticket:

- The Dockerfile you are using to build the image locally
- The URL of the Pipelines build that produces the Docker image with the issue

If you have any questions, please feel free to let me know.

Kind regards,
Theodora

[AZZ]

Thank you @Theodora Boudale . Will be on outlook for that email.

[Theodora Boudale]

Hi @AZZ,

I created the support ticket for you yesterday, and I see that you created a ticket yourself as well; I'm not sure if you received an email about the ticket I created for you.

This is the ticket I created (no one other than you and Atlassian can view its content):

• https://support.atlassian.com/requests/BBS-199063/

Please let me know if you can access it; I let my colleagues know about the duplicate tickets.

Kind regards,
Theodora