ci: fix multisite test

We just need to wait a little for the object to be replicated to the
other gateway. A simple retry solves this.

Closes: #8671
Signed-off-by: Sébastien Han <seb@redhat.com>

tests/scripts/github-action-helper.sh

@@ -253,7 +253,12 @@ function write_object_to_cluster1_read_from_cluster2() {
   s3cmd -v -d --config=s3cfg --access_key=${ACCESS_KEY} --secret_key=${SECRET_KEY} --host=${CLUSTER_1_IP_ADDR} mb s3://bkt
   s3cmd -v -d --config=s3cfg --access_key=${ACCESS_KEY} --secret_key=${SECRET_KEY} --host=${CLUSTER_1_IP_ADDR} put ./1M.dat s3://bkt
   CLUSTER_2_IP_ADDR=$(kubectl -n rook-ceph-secondary get svc rook-ceph-rgw-zone-b-multisite-store -o jsonpath="{.spec.clusterIP}")
-  s3cmd -v -d --config=s3cfg --access_key=${ACCESS_KEY} --secret_key=${SECRET_KEY} --host=${CLUSTER_2_IP_ADDR} get s3://bkt/1M.dat 1M-get.dat --force
+  timeout 60 bash <<EOF
+until s3cmd -v -d --config=s3cfg --access_key=${ACCESS_KEY} --secret_key=${SECRET_KEY} --host=${CLUSTER_2_IP_ADDR} get s3://bkt/1M.dat 1M-get.dat --force; do
+  echo "waiting for object to be replicated"
+  sleep 5
+done
+EOF
   diff 1M.dat 1M-get.dat
 }