diff --git a/.github/workflows/CRAN-R-CMD-check.yaml b/.github/workflows/CRAN-R-CMD-check.yaml
index cae83ca..2477d83 100644
--- a/.github/workflows/CRAN-R-CMD-check.yaml
+++ b/.github/workflows/CRAN-R-CMD-check.yaml
@@ -6,9 +6,9 @@
 # usethis::use_github_action("check-standard") will install it.
 on:
   push:
-    branches: [main, master]
+    branches: main
   pull_request:
-    branches: [main, master]
+    branches: main
   schedule:
     # runs tests every day at 1am EST
     - cron:  '0 5 * * *'
diff --git a/.github/workflows/GH-R-CMD-check.yaml b/.github/workflows/GH-R-CMD-check.yaml
index a23c081..806a3da 100644
--- a/.github/workflows/GH-R-CMD-check.yaml
+++ b/.github/workflows/GH-R-CMD-check.yaml
@@ -6,9 +6,9 @@
 # usethis::use_github_action("check-standard") will install it.
 on:
   push:
-    branches: [main, master]
+    branches: main
   pull_request:
-    branches: [main, master]
+    branches: main
   schedule:
     # runs tests every day at 1am EST
     - cron:  '0 5 * * *'
diff --git a/.github/workflows/spark-R-CMD-check.yaml b/.github/workflows/spark-R-CMD-check.yaml
index b377f21..f4d0281 100644
--- a/.github/workflows/spark-R-CMD-check.yaml
+++ b/.github/workflows/spark-R-CMD-check.yaml
@@ -6,9 +6,9 @@
 # usethis::use_github_action("check-standard") will install it.
 on:
   push:
-    branches: [main, master]
+    branches: main
   pull_request:
-    branches: [main, master]
+    branches: main
   schedule:
     # runs tests every day at 1am EST
     - cron:  '0 5 * * *'
@@ -26,13 +26,14 @@ jobs:
       fail-fast: false
       matrix:
         config:
-          - {os: ubuntu-latest, r: 'release'}
+          - {os: ubuntu-latest, r: 'release', spark: '4.0.1'}
 
     env:
       R_REMOTES_NO_ERRORS_FROM_WARNINGS: true
       GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
       R_COMPILE_AND_INSTALL_PACKAGES: 'always'
       R_KEEP_PKG_SOURCE: yes
+      SPARK_VERSION: ${{ matrix.config.spark }}
 
     steps:
       - uses: actions/checkout@v3
@@ -52,12 +53,14 @@ jobs:
 
       - uses: r-lib/actions/setup-r-dependencies@v2
         with:
-          extra-packages: any::rcmdcheck
+          extra-packages: |
+            any::rcmdcheck
+            any::devtools
           needs: check
 
       - name: Install Spark
         run: |
-          try(sparklyr::spark_install(version = "3.4.0", verbose = TRUE), silent = TRUE)
+          sparklyr::spark_install(version = Sys.getenv("SPARK_VERSION"))
         shell: Rscript {0}
 
       - name: Install devel versions
@@ -65,7 +68,10 @@ jobs:
           try(pak::pkg_install("tidymodels/parsnip"))
         shell: Rscript {0}
 
-      - uses: r-lib/actions/check-r-package@v2
+      - name: Run Spark tests
+        run: |
+          devtools::test(filter = "spark", stop_on_failure = TRUE)
+        shell: Rscript {0}
 
       - name: Notify slack fail
         if: failure() && github.event_name == 'schedule'
diff --git a/tests/testthat/helper-objects.R b/tests/testthat/helper-objects.R
index 3572f0b..bf3468d 100644
--- a/tests/testthat/helper-objects.R
+++ b/tests/testthat/helper-objects.R
@@ -1,6 +1,10 @@
 library(modeldata)
 library(parsnip)
 
+# Test tracker variables
+.env_tests <- new.env()
+.env_tests$spark_connection <- NULL
+
 ## -----------------------------------------------------------------------------
 
 data("wa_churn")
@@ -13,13 +17,34 @@ ctrl <- control_parsnip(verbosity = 1, catch = FALSE)
 caught_ctrl <- control_parsnip(verbosity = 1, catch = TRUE)
 quiet_ctrl <- control_parsnip(verbosity = 0, catch = TRUE)
 
-run_glmnet <- utils::compareVersion('3.6.0', as.character(getRversion())) > 0
+run_glmnet <- utils::compareVersion("3.6.0", as.character(getRversion())) > 0
 
-## -----------------------------------------------------------------------------
+## --------------------------- Spark -------------------------------------------
+
+skip_if_no_spark <- function() {
+  skip <- NULL
+  test_spark <- as.logical(Sys.getenv("TEST_SPARK", unset = TRUE))
+  if (!isTRUE(test_spark)) {
+    skip <- TRUE
+  }
+  if (is.null(skip)) {
+    if (spark_not_installed()) {
+      skip <- TRUE
+    } else {
+      if (is.null(spark_test_connection())) {
+        skip <- TRUE
+      }
+    }
+  }
+  if (skip %||% FALSE) {
+    skip("Skipping Spark related tests")
+  } else {
+    return(invisible())
+  }
+}
 
 spark_not_installed <- function() {
   need_install <- purrr:::quietly(sparklyr::spark_install_find)()
-
   if (inherits(need_install, "try-error")) {
     need_install <- TRUE
   } else {
@@ -28,6 +53,19 @@ spark_not_installed <- function() {
   need_install
 }
 
+spark_test_connection <- function() {
+  suppressPackageStartupMessages(library(sparklyr))
+  if (is.null(.env_tests$spark_connection)) {
+    version <- Sys.getenv("SPARK_VERSION", unset = "3.5.7")
+    sc <- try(sparklyr::spark_connect("local", version = version), silent = TRUE)
+    if (inherits(sc, "try-error")) {
+      return(NULL)
+    }
+    .env_tests$spark_connection <- sc
+  }
+  .env_tests$spark_connection
+}
+
 # ------------------------------------------------------------------------------
 
 expect_ptype <- function(x, ptype) {
diff --git a/tests/testthat/test-parsnip-case-weights.R b/tests/testthat/test-parsnip-case-weights.R
index 2508dcf..1336caa 100644
--- a/tests/testthat/test-parsnip-case-weights.R
+++ b/tests/testthat/test-parsnip-case-weights.R
@@ -4,11 +4,9 @@ skip_if_not_installed("hardhat", minimum_version = "1.2.0")
 skip_if_not_installed("yardstick", minimum_version = "1.0.0")
 skip_if_not_installed("workflows", minimum_version = "1.0.0")
 skip_if_not_installed("recipes", minimum_version = "1.0.0")
-skip_if_not_installed("sparklyr", minimum_version = "1.9.1.9000")
 
 # load all extension packages to register the engines
 library(parsnip)
-suppressPackageStartupMessages(library(sparklyr))
 
 # boosted trees -----------------------------------------------------------
 
@@ -184,53 +182,6 @@ test_that('linear_reg - glmnet case weights', {
   expect_equal(sub_fit$fit$beta, wt_fit$fit$beta)
 })
 
-test_that('linear_reg - spark case weights', {
-  skip_if_not_installed("sparklyr")
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
-
-  dat <- make_mtcars_wts()
-
-  mtcars_wts <- copy_to(
-    sc,
-    mtcars %>% mutate(wts = as.double(dat$wts)),
-    "dat_wts",
-    overwrite = TRUE
-  )
-
-  mtcars_subset <- copy_to(
-    sc,
-    dat$subset,
-    "mtcars_subset",
-    overwrite = TRUE
-  )
-
-  expect_error(
-    {
-      wt_fit <-
-        linear_reg() %>%
-        set_engine("spark") %>%
-        fit(
-          mpg ~ . - wts,
-          data = mtcars_wts,
-          case_weights = "wts"
-        )
-    },
-    regexp = NA
-  )
-
-  sub_fit <-
-    linear_reg() %>%
-    set_engine("spark") %>%
-    fit(mpg ~ ., data = mtcars_subset)
-
-  expect_equal(coef(sub_fit$fit), coef(wt_fit$fit))
-
-  spark_disconnect_all()
-})
-
 
 # logistic_reg ------------------------------------------------------------
 
@@ -296,54 +247,6 @@ test_that('logistic_reg - stan case weights', {
   expect_snapshot(print(wt_fit$fit$call))
 })
 
-test_that('logistic_reg - spark case weights', {
-  skip_if_not_installed("sparklyr")
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
-
-  dat <- make_two_class_wts()
-
-  two_class_dat_wts <- copy_to(
-    sc,
-    two_class_dat %>% mutate(wts = as.double(dat$wts)),
-    "two_class_dat_wts",
-    overwrite = TRUE
-  )
-
-  dat_subset <- copy_to(
-    sc,
-    dat$subset,
-    "dat_subset",
-    overwrite = TRUE
-  )
-
-  expect_error(
-    {
-      wt_fit <-
-        logistic_reg() %>%
-        set_engine("spark") %>%
-        fit(
-          Class ~ . - wts,
-          data = two_class_dat_wts,
-          case_weights = "wts"
-        )
-    },
-    regexp = NA
-  )
-
-  sub_fit <-
-    logistic_reg() %>%
-    set_engine("spark") %>%
-    fit(Class ~ ., data = dat_subset)
-
-  expect_equal(coef(sub_fit$fit), coef(wt_fit$fit))
-
-  spark_disconnect_all()
-})
-
-
 # mars --------------------------------------------------------------------
 
 test_that('mars - earth case weights', {
@@ -407,54 +310,6 @@ test_that('multinom_reg - glmnet case weights', {
   expect_equal(sub_fit$fit$beta, wt_fit$fit$beta)
 })
 
-test_that('multinom_reg - spark case weights', {
-  skip_if_not_installed("sparklyr")
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
-
-  dat <- make_penguin_wts()
-
-  penguin_wts <- copy_to(
-    sc,
-    penguins[complete.cases(penguins), ] %>% mutate(wts = as.double(dat$wts)),
-    "penguin_wts",
-    overwrite = TRUE
-  )
-
-  penguin_subset <- copy_to(
-    sc,
-    dat$subset,
-    "penguin_subset",
-    overwrite = TRUE
-  )
-
-  expect_error(
-    {
-      wt_fit <-
-        multinom_reg() %>%
-        set_engine("spark") %>%
-        fit(
-          island ~ . - wts,
-          data = penguin_wts,
-          case_weights = "wts"
-        )
-    },
-    regexp = NA
-  )
-
-  sub_fit <-
-    multinom_reg() %>%
-    set_engine("spark") %>%
-    fit(island ~ ., data = penguin_subset)
-
-  expect_equal(coef(sub_fit$fit), coef(wt_fit$fit))
-
-  spark_disconnect_all()
-})
-
-
 # rand_forest -------------------------------------------------------------
 
 test_that('rand_forest - ranger case weights', {
diff --git a/tests/testthat/test-spark-boost-tree.R b/tests/testthat/test-spark-boost-tree.R
index 8087c30..80e3ce5 100644
--- a/tests/testthat/test-spark-boost-tree.R
+++ b/tests/testthat/test-spark-boost-tree.R
@@ -1,6 +1,4 @@
-## Skip entire file is Spark is not installed
-skip_if(spark_not_installed())
-skip_if_not_installed("sparklyr", minimum_version = "1.9.1.9000")
+skip_if_no_spark()
 
 library(testthat)
 library(parsnip)
@@ -23,12 +21,7 @@ test_that('Reminder to check for CRAN sparklyr > 1.9.1.9000', {
 
 
 test_that('spark execution', {
-  skip_if_not_installed("sparklyr")
-  suppressPackageStartupMessages(library(sparklyr))
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
+  sc <- spark_test_connection()
 
   hpc_bt_tr <- copy_to(sc, hpc[-(1:4), ], "hpc_bt_tr", overwrite = TRUE)
   hpc_bt_te <- copy_to(sc, hpc[1:4, -1], "hpc_bt_te", overwrite = TRUE)
@@ -206,5 +199,4 @@ test_that('spark execution', {
     as.data.frame(spark_class_dup_classprob)
   )
 
-  spark_disconnect_all()
 })
diff --git a/tests/testthat/test-spark-case-weights.R b/tests/testthat/test-spark-case-weights.R
new file mode 100644
index 0000000..ce69180
--- /dev/null
+++ b/tests/testthat/test-spark-case-weights.R
@@ -0,0 +1,123 @@
+skip_if_no_spark()
+
+test_that("linear_reg - spark case weights", {
+  sc <- spark_test_connection()
+  dat <- make_mtcars_wts()
+
+  mtcars_wts <- copy_to(
+    sc,
+    mtcars %>% mutate(wts = as.double(dat$wts)),
+    "dat_wts",
+    overwrite = TRUE
+  )
+
+  mtcars_subset <- copy_to(
+    sc,
+    dat$subset,
+    "mtcars_subset",
+    overwrite = TRUE
+  )
+
+  expect_error(
+    {
+      wt_fit <-
+        linear_reg() %>%
+        set_engine("spark") %>%
+        fit(
+          mpg ~ . - wts,
+          data = mtcars_wts,
+          case_weights = "wts"
+        )
+    },
+    regexp = NA
+  )
+
+  sub_fit <-
+    linear_reg() %>%
+    set_engine("spark") %>%
+    fit(mpg ~ ., data = mtcars_subset)
+
+  expect_equal(coef(sub_fit$fit), coef(wt_fit$fit))
+})
+
+test_that("logistic_reg - spark case weights", {
+  sc <- spark_test_connection()
+
+  dat <- make_two_class_wts()
+
+  two_class_dat_wts <- copy_to(
+    sc,
+    two_class_dat %>% mutate(wts = as.double(dat$wts)),
+    "two_class_dat_wts",
+    overwrite = TRUE
+  )
+
+  dat_subset <- copy_to(
+    sc,
+    dat$subset,
+    "dat_subset",
+    overwrite = TRUE
+  )
+
+  expect_error(
+    {
+      wt_fit <-
+        logistic_reg() %>%
+        set_engine("spark") %>%
+        fit(
+          Class ~ . - wts,
+          data = two_class_dat_wts,
+          case_weights = "wts"
+        )
+    },
+    regexp = NA
+  )
+
+  sub_fit <-
+    logistic_reg() %>%
+    set_engine("spark") %>%
+    fit(Class ~ ., data = dat_subset)
+
+  expect_equal(coef(sub_fit$fit), coef(wt_fit$fit))
+})
+
+test_that("multinom_reg - spark case weights", {
+  sc <- spark_test_connection()
+
+  dat <- make_penguin_wts()
+
+  penguin_wts <- copy_to(
+    sc,
+    penguins[complete.cases(penguins), ] %>% mutate(wts = as.double(dat$wts)),
+    "penguin_wts",
+    overwrite = TRUE
+  )
+
+  penguin_subset <- copy_to(
+    sc,
+    dat$subset,
+    "penguin_subset",
+    overwrite = TRUE
+  )
+
+  expect_error(
+    {
+      wt_fit <-
+        multinom_reg() %>%
+        set_engine("spark") %>%
+        fit(
+          island ~ . - wts,
+          data = penguin_wts,
+          case_weights = "wts"
+        )
+    },
+    regexp = NA
+  )
+
+  sub_fit <-
+    multinom_reg() %>%
+    set_engine("spark") %>%
+    fit(island ~ ., data = penguin_subset)
+
+  expect_equal(coef(sub_fit$fit), coef(wt_fit$fit))
+})
diff --git a/tests/testthat/test-spark-data-descriptors.R b/tests/testthat/test-spark-data-descriptors.R
index d441797..e224f42 100644
--- a/tests/testthat/test-spark-data-descriptors.R
+++ b/tests/testthat/test-spark-data-descriptors.R
@@ -1,6 +1,4 @@
-## Skip entire file is Spark is not installed
-skip_if(spark_not_installed())
-skip_if_not_installed("sparklyr", minimum_version = "1.9.1.9000")
+skip_if_no_spark()
 
 library(testthat)
 library(parsnip)
@@ -39,14 +37,7 @@ class_tab <- table(hpc$class, dnn = NULL)
 # ------------------------------------------------------------------------------
 
 test_that("spark descriptor", {
-  skip_if_not_installed("sparklyr")
-
-  suppressPackageStartupMessages(library(sparklyr))
-  library(dplyr)
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
+  sc <- spark_test_connection()
 
   npk_descr <- copy_to(sc, npk[, 1:4], "npk_descr", overwrite = TRUE)
   hpc_descr <- copy_to(sc, hpc, "hpc_descr", overwrite = TRUE)
@@ -88,5 +79,4 @@ test_that("spark descriptor", {
     eval_descrs2(parsnip:::get_descr_form(K ~ ., data = npk_descr)),
     ignore_attr = TRUE
   )
-  spark_disconnect_all()
 })
diff --git a/tests/testthat/test-spark-linear-reg.R b/tests/testthat/test-spark-linear-reg.R
index 2f615c9..4a04d83 100644
--- a/tests/testthat/test-spark-linear-reg.R
+++ b/tests/testthat/test-spark-linear-reg.R
@@ -1,6 +1,4 @@
-## Skip entire file is Spark is not installed
-skip_if(spark_not_installed())
-skip_if_not_installed("sparklyr", minimum_version = "1.9.1.9000")
+skip_if_no_spark()
 
 library(testthat)
 library(parsnip)
@@ -13,13 +11,7 @@ hpc <- hpc_data[1:150, c(2:5, 8)]
 # ------------------------------------------------------------------------------
 
 test_that('spark execution', {
-  skip_if_not_installed("sparklyr")
-
-  suppressPackageStartupMessages(library(sparklyr))
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
+  sc <- spark_test_connection()
 
   hpc_linreg_tr <- copy_to(sc, hpc[-(1:4), ], "hpc_linreg_tr", overwrite = TRUE)
   hpc_linreg_te <- copy_to(sc, hpc[1:4, -1], "hpc_linreg_te", overwrite = TRUE)
@@ -53,6 +45,4 @@ test_that('spark execution', {
 
   expect_equal(as.data.frame(spark_pred)$pred, lm_pred)
   expect_equal(as.data.frame(spark_pred_num)$pred, lm_pred)
-
-  spark_disconnect_all()
 })
diff --git a/tests/testthat/test-spark-logistic-reg.R b/tests/testthat/test-spark-logistic-reg.R
index 492245e..989caea 100644
--- a/tests/testthat/test-spark-logistic-reg.R
+++ b/tests/testthat/test-spark-logistic-reg.R
@@ -1,6 +1,4 @@
-## Skip entire file is Spark is not installed
-skip_if(spark_not_installed())
-skip_if_not_installed("sparklyr", minimum_version = "1.9.1.9000")
+skip_if_no_spark()
 
 library(testthat)
 library(parsnip)
@@ -13,13 +11,7 @@ hpc <- hpc_data[1:150, c(2:5, 8)]
 # ------------------------------------------------------------------------------
 
 test_that('spark execution', {
-  skip_if_not_installed("sparklyr")
-
-  suppressPackageStartupMessages(library(sparklyr))
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
+  sc <- spark_test_connection()
 
   churn_logit_tr <- copy_to(
     sc,
@@ -97,6 +89,4 @@ test_that('spark execution', {
     as.data.frame(spark_class_prob_classprob),
     ignore_attr = TRUE
   )
-
-  spark_disconnect_all()
 })
diff --git a/tests/testthat/test-spark-multinom-reg.R b/tests/testthat/test-spark-multinom-reg.R
index 97541d6..68953f0 100644
--- a/tests/testthat/test-spark-multinom-reg.R
+++ b/tests/testthat/test-spark-multinom-reg.R
@@ -1,6 +1,4 @@
-## Skip entire file is Spark is not installed
-skip_if(spark_not_installed())
-skip_if_not_installed("sparklyr", minimum_version = "1.9.1.9000")
+skip_if_no_spark()
 
 library(testthat)
 library(parsnip)
@@ -13,13 +11,7 @@ hpc <- hpc_data[1:150, c(2:5, 8)]
 # ------------------------------------------------------------------------------
 
 test_that('spark execution', {
-  skip_if_not_installed("sparklyr")
-
-  library(sparklyr)
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
+  sc <- spark_test_connection()
 
   hpc_rows <- c(1, 51, 101)
   hpc_tr <- copy_to(sc, hpc[-hpc_rows, ], "hpc_tr", overwrite = TRUE)
@@ -80,5 +72,4 @@ test_that('spark execution', {
     ignore_attr = TRUE
   )
 
-  spark_disconnect_all()
 })
diff --git a/tests/testthat/test-spark-rand-forest.R b/tests/testthat/test-spark-rand-forest.R
index c2cb907..60384bd 100644
--- a/tests/testthat/test-spark-rand-forest.R
+++ b/tests/testthat/test-spark-rand-forest.R
@@ -1,6 +1,4 @@
-## Skip entire file is Spark is not installed
-skip_if(spark_not_installed())
-skip_if_not_installed("sparklyr", minimum_version = "1.9.1.9000")
+skip_if_no_spark()
 
 library(testthat)
 library(parsnip)
@@ -13,13 +11,7 @@ hpc <- hpc_data[1:150, c(2:5, 8)]
 # ------------------------------------------------------------------------------
 
 test_that('spark execution', {
-  skip_if_not_installed("sparklyr")
-
-  suppressPackageStartupMessages(library(sparklyr))
-
-  sc <- try(spark_connect(master = "local"), silent = TRUE)
-
-  skip_if(inherits(sc, "try-error"))
+  sc <- spark_test_connection()
 
   hpc_rf_tr <- copy_to(sc, hpc[-(1:4), ], "hpc_rf_tr", overwrite = TRUE)
   hpc_rf_te <- copy_to(sc, hpc[1:4, -1], "hpc_rf_te", overwrite = TRUE)
@@ -185,6 +177,4 @@ test_that('spark execution', {
     as.data.frame(spark_class_prob_classprob),
     as.data.frame(spark_class_dup_classprob)
   )
-
-  spark_disconnect_all()
 })