AZP/AWS: EFA tests #10494
AZP/AWS: EFA tests #10494
Conversation
Alexey-Rivkin
force-pushed
the
topic/efa_aws_tests
branch
29 times, most recently
from
12da037 to
5a9eac6
Compare
Alexey-Rivkin
force-pushed
the
topic/efa_aws_tests
branch
19 times, most recently
from
e3159d3 to
9fa1f06
Compare
Alexey-Rivkin
force-pushed
the
topic/efa_aws_tests
branch
from
9fa1f06 to
afbfb21
Compare
|
test |
buildlib/pr/aws/efa_aws.yml
Outdated
| RUN_TESTS: yes | ||
| TEST_PERF: 0 | ||
| VALGRIND_CHECK: no | ||
| CMD: "yum groupinstall -y 'Development Tools' && yum install -y fuse-devel iproute hostname strace git wget environment-modules autoconf libtool python3 python3-pip pkg-config libnl3-devel curl valgrind valgrind-devel rdma-core-devel libibverbs libibverbs-utils librdmacm librdmacm-utils && git clone https://github.com/openucx/ucx.git && cd ucx && git fetch origin pull/${PR_ID}/head:pr && git checkout pr && ./contrib/test_jenkins.sh" |
There was a problem hiding this comment.
Why are you always installing these tools? Why not create single image with all of preinstalled prerequisites?
There was a problem hiding this comment.
We evaluated creating a prebuilt image but opted against it.
Installation takes under 20 seconds, and maintaining an additional image adds overhead in management and documentation.
buildlib/pr/aws/efa_aws.yml
Outdated
| RUN_TESTS: yes | ||
| TEST_PERF: 0 | ||
| VALGRIND_CHECK: no | ||
| CMD: "yum groupinstall -y 'Development Tools' && yum install -y fuse-devel iproute hostname strace git wget environment-modules autoconf libtool python3 python3-pip pkg-config libnl3-devel curl valgrind valgrind-devel rdma-core-devel libibverbs libibverbs-utils librdmacm librdmacm-utils && git clone https://github.com/openucx/ucx.git && cd ucx && git fetch origin pull/${PR_ID}/head:pr && git checkout pr && ./contrib/test_jenkins.sh" |
There was a problem hiding this comment.
git clone https://github.com/openucx/ucx.git
why not shallow copy? this will speed up the checkout
There was a problem hiding this comment.
Good point, thanks!
AZP/AWS: Shallow clone
| clean: true | ||
| - bash: | | ||
| set -x | ||
| aws eks update-kubeconfig --name ucx-ci |
There was a problem hiding this comment.
Why do you need it here? The conf should be already in place
There was a problem hiding this comment.
A separate job spins another container, so the config is empty.
buildlib/pr/aws/efa_aws.yml
Outdated
| RUN_TESTS: yes | ||
| TEST_PERF: 0 | ||
| VALGRIND_CHECK: no | ||
| CMD: "yum groupinstall -y 'Development Tools' && yum install -y fuse-devel iproute hostname strace git wget environment-modules autoconf libtool python3 python3-pip pkg-config libnl3-devel curl valgrind valgrind-devel rdma-core-devel libibverbs libibverbs-utils librdmacm librdmacm-utils && git clone https://github.com/openucx/ucx.git && cd ucx && git fetch origin pull/${PR_ID}/head:pr && git checkout pr && ./contrib/test_jenkins.sh" |
There was a problem hiding this comment.
IMO it's worth moving the contents of CMD command to dedicated script
It'll be much more readable and if there will be changes one will be able to see diffs
There was a problem hiding this comment.
Any script would be part of the codebase, but the container starts as a blank slate.
It can only run code after installing packages and cloning the repo - which is most of this CMD.
There was a problem hiding this comment.
IMO moving CMD in to efa_aws_test.sh make more sense
There was a problem hiding this comment.
AZP/AWS: Define AWS_CMD in the shell script
Good idea, thank you.
Alexey-Rivkin
force-pushed
the
topic/efa_aws_tests
branch
from
624f5bb to
e9a8552
Compare
Alexey-Rivkin
added
Ready for Review
and removed
WIP-DNM
What?
Add EFA tests in AWS.
Why?
Run tests on the real EFA environment.
How?
Use AWS batch to start a predefined AWS job.
Note
The scope of tests to run in AWS is TBD. Until then, the job will issue a warning instead of failing.