Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[v1.22.x] contrib/intel/jenkins: cherrypick CI changes from #10642 #10645

Merged
merged 5 commits into from
Dec 19, 2024
Merged

[v1.22.x] contrib/intel/jenkins: cherrypick CI changes from #10642 #10645

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
409094b
contrib/intel/jenkins: Update slurm partitions for new head node
zachdworkin Dec 12, 2024
ba2fac9
contrib/intel/jenkins: Uplevel pre-build directory
zachdworkin Dec 5, 2024
9f22685
contrib/intel/jenkins: Force Cleanup in Post
zachdworkin Dec 17, 2024
8fce5bb
contrib/intel/jenkins: Cleanup trailing whitespace
zachdworkin Dec 17, 2024
78c8cdb
contrib/intel/jenkins: Re-order stages to put slow ones first
zachdworkin Dec 17, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
119 changes: 63 additions & 56 deletions contrib/intel/jenkins/Jenkinsfile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,6 @@ def run_python(version, command, output=null) {
}

def slurm_batch(partition, node_num, output, command) {

try {
sh """sbatch --partition=${partition} -N ${node_num} \
--wait -o ${output} --open-mode=append \
Expand Down Expand Up @@ -105,9 +104,16 @@ def run_ci(stage_name, config_name) {
)
}

def build_ci(config_name) {
sh """source ${CI_LOCATION}/${env.CI_MODULE}/venv/bin/activate;\
python run.py \
--output=${env.CUSTOM_WORKSPACE}/pre-build \
--job=${config_name}
"""
}

def gather_logs(cluster, key, dest, source) {
def address = "${env.USER}@${cluster}"

try {
sh "scp -r -i ${key} ${address}:${source}/* ${dest}/"
} catch (Exception e) {
Expand Down Expand Up @@ -443,7 +449,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_water.json")
build_ci("pr_build_water.json")
}
}
}
Expand All @@ -452,7 +458,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_grass.json")
build_ci("pr_build_grass.json")
}
}
}
Expand All @@ -461,7 +467,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_electric.json")
build_ci("pr_build_electric.json")
}
}
}
Expand All @@ -470,7 +476,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_cyndaquil.json")
build_ci("pr_build_cyndaquil.json")
}
}
}
Expand All @@ -479,7 +485,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_quilava.json")
build_ci("pr_build_quilava.json")
}
}
}
Expand All @@ -501,7 +507,7 @@ pipeline {
}
bootstrap_ci()
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_daos.json")
build_ci("pr_build_daos.json")
}
}
}
Expand All @@ -524,7 +530,7 @@ pipeline {
}
bootstrap_ci()
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_fire.json")
build_ci("pr_build_fire.json")
}
}
}
Expand All @@ -538,7 +544,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_shmem_water.json")
build_ci("pr_build_shmem_water.json")
}
}
}
Expand All @@ -547,7 +553,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_shmem_grass.json")
build_ci("pr_build_shmem_grass.json")
}
}
}
Expand All @@ -556,7 +562,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_ompi_water.json")
build_ci("pr_build_ompi_water.json")
}
}
}
Expand All @@ -565,7 +571,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
run_ci("pre-build", "pr_build_ompi_grass.json")
build_ci("pr_build_ompi_grass.json")
}
}
}
Expand All @@ -574,7 +580,7 @@ pipeline {
steps {
script {
dir (CI_LOCATION) {
slurm_batch("totodile", "1",
slurm_batch("water", "1",
"${env.LOG_DIR}/build_mpich_water_log",
"""python$PYTHON_VERSION ${RUN_LOCATION}/build.py \
--build_item=mpich --build_hw=water"""
Expand All @@ -601,6 +607,41 @@ pipeline {
stage('parallel-tests') {
when { equals expected: true, actual: DO_RUN }
parallel {
stage('mpichtestsuite-tcp') {
steps {
script {
dir (RUN_LOCATION) {
def providers = [['tcp', null]]
def MPIS = ["mpich"]
if (env.WEEKLY.toBoolean()) {
MPIS = ["impi", "mpich"]
}
for (def mpi in MPIS) {
run_middleware(providers, "mpichtestsuite", "mpichtestsuite",
"grass", "bulbasaur,ivysaur", "2", "${mpi}")
}
}
}
}
}
stage('mpichtestsuite-verbs') {
steps {
script {
dir (RUN_LOCATION) {
def providers = [["verbs","rxm"]]
def MPIS = ["mpich"]
if (env.WEEKLY.toBoolean()) {
MPIS = ["impi", "mpich"]
}
for (def mpi in MPIS) {
run_middleware(providers, "mpichtestsuite", "mpichtestsuite",
"water", "squirtle,wartortle,articuno", "2",
"${mpi}")
}
}
}
}
}
stage ('CI_mpi_verbs-rxm_imb') {
steps {
script {
Expand Down Expand Up @@ -752,23 +793,6 @@ pipeline {
}
}
}
stage('mpichtestsuite') {
steps {
script {
dir (RUN_LOCATION) {
def providers = [['tcp', null], ["verbs","rxm"]]
def MPIS = ["mpich"]
if (env.WEEKLY.toBoolean()) {
MPIS = ["impi", "mpich"]
}
for (def mpi in MPIS) {
run_middleware(providers, "mpichtestsuite", "mpichtestsuite",
"water", "totodile", "2", "${mpi}")
}
}
}
}
}
stage('SHMEM_grass') {
steps {
script {
Expand All @@ -794,9 +818,9 @@ pipeline {
script {
dir (RUN_LOCATION) {
run_middleware([["tcp", null],["sockets", null]],
"multinode_performance", "multinode", "grass", "grass", "2")
"multinode_performance", "multinode", "grass", "bulbasaur,ivysaur", "2")
run_middleware([["verbs", "rxm"]], "multinode_performance",
"multinode", "water", "totodile", "2")
"multinode", "water", "squirtle,wartortle,articuno", "2")
}
}
}
Expand All @@ -806,13 +830,15 @@ pipeline {
script {
dir (RUN_LOCATION) {
run_middleware([["verbs", null]], "oneCCL",
"oneccl", "water", "totodile", "2")
"oneccl", "water",
"squirtle,wartortle,articuno", "2")
run_middleware([["shm", null]], "oneCCL",
"oneccl", "grass", "bulbasaur", "1")
"oneccl", "grass", "bulbasaur,ivysaur", "1")
run_middleware([["psm3", null]], "oneCCL",
"oneccl", "water", "totodile", "2")
"oneccl", "water",
"squirtle,wartortle,articuno", "2")
run_middleware([["tcp", null]], "oneCCL",
"oneccl", "grass", "bulbasaur", "2")
"oneccl", "grass", "bulbasaur,ivysaur", "2")
run_middleware([["shm", null]], "oneCCL_DSA",
"oneccl", "electric", "pikachu", "1", null, null,
"""CCL_ATL_SHM=1 FI_SHM_DISABLE_CMA=1 \
Expand Down Expand Up @@ -969,25 +995,6 @@ pipeline {
summarize("all")
}
}
}
aborted {
node ('daos_head') {
dir ("${DELETE_LOCATION}/middlewares") { deleteDir() }
}
node ('ze') {
dir ("${DELETE_LOCATION}/middlewares") { deleteDir() }
}
sh "scancel --jobname=\"${SLURM_JOB_NAME}\""
dir ("${DELETE_LOCATION}/middlewares") { deleteDir() }
}
success {
script {
if (DO_RUN) {
CI_summarize(verbose=true)
summarize("all", verbose=true, release=false,
send_mail=env.WEEKLY.toBoolean())
}
}
node ('daos_head') {
dir("${env.WORKSPACE}") { deleteDir() }
dir("${env.WORKSPACE}@tmp") { deleteDir() }
Expand Down
Loading