From 26dfdc18f4244b3b2e6f8ee0afc7c4b3700da2da Mon Sep 17 00:00:00 2001
From: Watson Sato <wsato@redhat.com>
Date: Mon, 29 Jul 2024 15:06:43 +0200
Subject: [PATCH] Fetch the raw results in parallel

Move the commands extracting the raw-results to a function and run it in
the background.
We also wait for them all to be finished, to ensure the script doesn't
terminate while a pod is still extracting raw results.
---
 utils/must-gather/gather_compliance | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/utils/must-gather/gather_compliance b/utils/must-gather/gather_compliance
index e661f0bd5..c7d2b6d2f 100755
--- a/utils/must-gather/gather_compliance
+++ b/utils/must-gather/gather_compliance
@@ -19,13 +19,8 @@ do
   done
 done
 
-NAMESPACE="openshift-compliance"
-SCANS=$(${BIN} get scans -n ${NAMESPACE} -o jsonpath="{.items[*].metadata.name}")
-RAW_RESULTS_DIR=${LOGS_DIR}/${NAMESPACE}/raw-results/
-EXTRACT_PODS_DIR=${RAW_RESULTS_DIR}/extract-pods/
-mkdir -p ${EXTRACT_PODS_DIR}
-for SCAN in ${SCANS[@]}
-do
+function extract_raw_results(){
+    SCAN=$1
     mkdir ${RAW_RESULTS_DIR}/${SCAN}
     CLAIMNAME=$(${BIN} get pvc -n ${NAMESPACE} ${SCAN} -ojsonpath='{.metadata.name}')
     # The EXTRACT_POD_NAME needs to match the pod name in fetch-raw-results-pod-template.yaml
@@ -36,7 +31,19 @@ do
     ${BIN} wait -n ${NAMESPACE} --for=condition=Ready pod/${EXTRACT_POD_NAME}
     ${BIN} cp -n ${NAMESPACE} ${EXTRACT_POD_NAME}:/scan-results ${RAW_RESULTS_DIR}/${SCAN}
     ${BIN} delete pod -n ${NAMESPACE} ${EXTRACT_POD_NAME}
+}
+
+NAMESPACE="openshift-compliance"
+SCANS=$(${BIN} get scans -n ${NAMESPACE} -o jsonpath="{.items[*].metadata.name}")
+RAW_RESULTS_DIR=${LOGS_DIR}/${NAMESPACE}/raw-results/
+EXTRACT_PODS_DIR=${RAW_RESULTS_DIR}/extract-pods/
+mkdir -p ${EXTRACT_PODS_DIR}
+for SCAN in ${SCANS[@]}
+do
+    extract_raw_results ${SCAN} &
 done
+# Let's wait for all raw-results to be extracted
+wait
 
 # Pod logs, describes
 NAMESPACES=(openshift-compliance)