feat(etl): SOAK

Signed-off-by: Peter Mikus <peter.mikus@protonmail.ch>
Change-Id: Ic8cd4cc6de271ca39b2b9d4e07da2718772e9a38

2 files changed, 245 insertions, 171 deletions

diff --git a/fdio.infra.terraform/terraform-nomad-pyspark-etl/conf/nomad/etl-trending-soak.hcl.tftpl b/fdio.infra.terraform/terraform-nomad-pyspark-etl/conf/nomad/etl-trending-soak.hcl.tftpl
new file mode 100644
index 0000000000..6d77a898df
--- /dev/null
+++ b/fdio.infra.terraform/terraform-nomad-pyspark-etl/conf/nomad/etl-trending-soak.hcl.tftpl
@@ -0,0 +1,55 @@
+job "${job_name}" {
+  datacenters = "${datacenters}"
+  type        = "${type}"
+  periodic {
+    cron             = "${cron}"
+    prohibit_overlap = "${prohibit_overlap}"
+    time_zone        = "${time_zone}"
+  }
+  group "${job_name}" {
+    restart {
+      mode = "fail"
+    }
+    constraint {
+      attribute       = "$${attr.cpu.arch}"
+      operator        = "!="
+      value           = "arm64"
+    }
+    constraint {
+      attribute      = "$${node.class}"
+      value          = "builder"
+    }
+    task "${job_name}" {
+      artifact {
+        source      = "git::https://github.com/FDio/csit"
+        destination = "local/csit"
+      }
+      driver = "docker"
+      config {
+        image   = "${image}"
+        command = "gluesparksubmit"
+        args = [
+          "--driver-memory", "30g",
+          "--executor-memory", "30g",
+          "--executor-cores", "2",
+          "--master", "local[2]",
+          "trending_soak.py"
+        ]
+        work_dir = "/local/csit/csit.infra.etl"
+      }
+      env {
+        AWS_ACCESS_KEY_ID         = "${aws_access_key_id}"
+        AWS_SECRET_ACCESS_KEY     = "${aws_secret_access_key}"
+        AWS_DEFAULT_REGION        = "${aws_default_region}"
+        OUT_AWS_ACCESS_KEY_ID     = "${out_aws_access_key_id}"
+        OUT_AWS_SECRET_ACCESS_KEY = "${out_aws_secret_access_key}"
+        OUT_AWS_DEFAULT_REGION    = "${out_aws_default_region}"
+        ${ envs }
+      }
+      resources {
+        cpu    = ${cpu}
+        memory = ${memory}
+      }
+    }
+  }
+}
diff --git a/fdio.infra.terraform/terraform-nomad-pyspark-etl/fdio/main.tf b/fdio.infra.terraform/terraform-nomad-pyspark-etl/fdio/main.tf
index 6b5edb79b7..aac81d9b78 100644
--- a/fdio.infra.terraform/terraform-nomad-pyspark-etl/fdio/main.tf
+++ b/fdio.infra.terraform/terraform-nomad-pyspark-etl/fdio/main.tf
@@ -76,7 +76,7 @@ module "etl-trending-ndrpdr" {
   memory                    = 60000
 }
 
-module "etl-iterative-hoststack-rls2402" {
+module "etl-trending-soak" {
   providers = {
     nomad = nomad.yul1
   }
@@ -90,175 +90,194 @@ module "etl-iterative-hoststack-rls2402" {
   out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
   cron                      = "0 30 0 * * * *"
   datacenters               = ["yul1"]
-  job_name                  = "etl-iterative-hoststack-rls2402"
-}
-
-module "etl-iterative-mrr-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-iterative-mrr-rls2402"
-}
-
-module "etl-iterative-ndrpdr-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-iterative-ndrpdr-rls2402"
-}
-
-module "etl-iterative-reconf-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-iterative-reconf-rls2402"
-}
-
-module "etl-iterative-soak-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-iterative-soak-rls2402"
-}
-
-module "etl-coverage-device-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-coverage-device-rls2402"
-}
-
-module "etl-coverage-hoststack-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-coverage-hoststack-rls2402"
-}
-
-module "etl-coverage-mrr-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-coverage-mrr-rls2402"
-}
-
-module "etl-coverage-ndrpdr-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-coverage-ndrpdr-rls2402"
-}
-
-module "etl-coverage-reconf-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-coverage-reconf-rls2402"
+  job_name                  = "etl-trending-soak"
+  memory                    = 60000
 }
 
-module "etl-coverage-soak-rls2402" {
-  providers = {
-    nomad = nomad.yul1
-  }
-  source = "../"
-
-  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
-  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
-  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
-  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
-  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
-  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
-  cron                      = "0 30 0 * * * *"
-  datacenters               = ["yul1"]
-  job_name                  = "etl-coverage-soak-rls2402"
-}
+#module "etl-iterative-hoststack-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-iterative-hoststack-rls2402"
+#}
+#
+#module "etl-iterative-mrr-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-iterative-mrr-rls2402"
+#}
+#
+#module "etl-iterative-ndrpdr-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-iterative-ndrpdr-rls2402"
+#}
+#
+#module "etl-iterative-reconf-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-iterative-reconf-rls2402"
+#}
+#
+#module "etl-iterative-soak-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-iterative-soak-rls2402"
+#}
+#
+#module "etl-coverage-device-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-coverage-device-rls2402"
+#}
+#
+#module "etl-coverage-hoststack-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-coverage-hoststack-rls2402"
+#}
+#
+#module "etl-coverage-mrr-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-coverage-mrr-rls2402"
+#}
+#
+#module "etl-coverage-ndrpdr-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-coverage-ndrpdr-rls2402"
+#}
+#
+#module "etl-coverage-reconf-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-coverage-reconf-rls2402"
+#}
+#
+#module "etl-coverage-soak-rls2402" {
+#  providers = {
+#    nomad = nomad.yul1
+#  }
+#  source = "../"
+#
+#  aws_access_key_id         = data.vault_generic_secret.fdio_logs.data["access_key"]
+#  aws_secret_access_key     = data.vault_generic_secret.fdio_logs.data["secret_key"]
+#  aws_default_region        = data.vault_generic_secret.fdio_logs.data["region"]
+#  out_aws_access_key_id     = data.vault_generic_secret.fdio_docs.data["access_key"]
+#  out_aws_secret_access_key = data.vault_generic_secret.fdio_docs.data["secret_key"]
+#  out_aws_default_region    = data.vault_generic_secret.fdio_docs.data["region"]
+#  cron                      = "0 30 0 * * * *"
+#  datacenters               = ["yul1"]
+#  job_name                  = "etl-coverage-soak-rls2402"
+#}
+#
\ No newline at end of file