refactor(tests): always check ssh output

Ssh connection failure should be considered an error by default. In order
to unify handling of this case, move check for non zero return code
into the function itself. Because of this `check_output` method is
not needed anymore and was removed.

Signed-off-by: Egor Lazarchuk <[email protected]>

Author: ShadowCurse

tests/framework/microvm.py

@@ -1003,14 +1003,13 @@ def thread_backtraces(self):
                 )
         return "\n".join(backtraces)
 
-    def _dump_debug_information(self, exc: Exception):
+    def _dump_debug_information(self):
         """
         Dumps debug information about this microvm
 
         Used for example when running a command inside the guest via `SSHConnection.check_output` fails.
         """
         print(
-            f"Failure executing command via SSH in microVM: {exc}\n\n"
             f"Firecracker logs:\n{self.log_data}\n"
             f"Thread backtraces:\n{self.thread_backtraces}"
         )

tests/framework/utils.py

@@ -350,8 +350,7 @@ def get_free_mem_ssh(ssh_connection):
     :param ssh_connection: connection to the guest
     :return: available mem column output of 'free'
     """
-    _, stdout, stderr = ssh_connection.run("cat /proc/meminfo | grep MemAvailable")
-    assert stderr == ""
+    _, stdout, _ = ssh_connection.run("cat /proc/meminfo | grep MemAvailable")
 
     # Split "MemAvailable:   123456 kB" and validate it
     meminfo_data = stdout.split()
@@ -625,8 +624,7 @@ def guest_run_fio_iteration(ssh_connection, iteration):
         --output /tmp/fio{} > /dev/null &""".format(
         iteration
     )
-    exit_code, _, stderr = ssh_connection.run(fio)
-    assert exit_code == 0, stderr
+    ssh_connection.run(fio)
 
 
 def check_filesystem(ssh_connection, disk_fmt, disk):

tests/framework/utils_iperf.py

@@ -119,7 +119,7 @@ def spawn_iperf3_client(self, client_idx, client_mode_flag):
             .build()
         )
 
-        return self._microvm.ssh.check_output(cmd).stdout
+        return self._microvm.ssh.run(cmd).stdout
 
     def host_command(self, port_offset):
         """Builds the command used for spawning an iperf3 server on the host"""

tests/framework/utils_vsock.py

@@ -100,7 +100,7 @@ def start_guest_echo_server(vm):
     Returns a UDS path to connect to the server.
     """
     cmd = f"nohup socat VSOCK-LISTEN:{ECHO_SERVER_PORT},backlog=128,reuseaddr,fork EXEC:'/bin/cat' > /dev/null 2>&1 &"
-    vm.ssh.check_output(cmd)
+    vm.ssh.run(cmd)
 
     # Give the server time to initialise
     time.sleep(1)
@@ -214,8 +214,7 @@ def _copy_vsock_data_to_guest(ssh_connection, blob_path, vm_blob_path, vsock_hel
     # Copy the data file and a vsock helper to the guest.
 
     cmd = "mkdir -p /tmp/vsock"
-    ecode, _, _ = ssh_connection.run(cmd)
-    assert ecode == 0, "Failed to set up tmpfs drive on the guest."
+    ssh_connection.run(cmd)
 
     ssh_connection.scp_put(vsock_helper, "/tmp/vsock_helper")
     ssh_connection.scp_put(blob_path, vm_blob_path)

tests/host_tools/network.py

@@ -58,13 +58,7 @@ def __init__(self, netns, ssh_key: Path, host, user, *, on_error=None):
         # _init_connection loops until it can connect to the guest
         # dumping debug state on every iteration is not useful or wanted, so
         # only dump it once if _all_ iterations fail.
-        try:
-            self._init_connection()
-        except Exception as exc:
-            if on_error:
-                on_error(exc)
-
-            raise
+        self._init_connection()
 
         self._on_error = on_error
 
@@ -79,7 +73,7 @@ def remote_path(self, path):
 
     def _scp(self, path1, path2, options):
         """Copy files to/from the VM using scp."""
-        self._exec(["scp", *options, path1, path2], check=True)
+        self._exec(["scp", *options, path1, path2])
 
     def scp_put(self, local_path, remote_path, recursive=False):
         """Copy files to the VM using scp."""
@@ -96,7 +90,7 @@ def scp_get(self, remote_path, local_path, recursive=False):
         self._scp(self.remote_path(remote_path), local_path, opts)
 
     @retry(
-        retry=retry_if_exception_type(ChildProcessError),
+        retry=retry_if_exception_type(AssertionError),
         wait=wait_fixed(0.5),
         stop=stop_after_attempt(20),
         reraise=True,
@@ -109,11 +103,12 @@ def _init_connection(self):
         We'll keep trying to execute a remote command that can't fail
         (`/bin/true`), until we get a successful (0) exit code.
         """
-        self.check_output("true", timeout=100, debug=True)
+        self.run("true", timeout=100, debug=True)
 
-    def run(self, cmd_string, timeout=None, *, check=False, debug=False):
+    def run(self, cmd_string, timeout=None, *, check=True, debug=False):
         """
         Execute the command passed as a string in the ssh context.
+        By default raises an exception on non-zero return code of remote command.
 
         If `debug` is set, pass `-vvv` to `ssh`. Note that this will clobber stderr.
         """
@@ -124,22 +119,29 @@ def run(self, cmd_string, timeout=None, *, check=False, debug=False):
 
         return self._exec(command, timeout, check=check)
 
-    def check_output(self, cmd_string, timeout=None, *, debug=False):
-        """Same as `run`, but raises an exception on non-zero return code of remote command"""
-        return self.run(cmd_string, timeout, check=True, debug=debug)
-
-    def _exec(self, cmd, timeout=None, check=False):
+    def _exec(self, cmd, timeout=None, check=True):
         """Private function that handles the ssh client invocation."""
         if self.netns is not None:
             cmd = ["ip", "netns", "exec", self.netns] + cmd
 
-        try:
-            return utils.run_cmd(cmd, check=check, timeout=timeout)
-        except Exception as exc:
+        rc, stdout, stderr = utils.run_cmd(cmd, timeout=timeout)
+
+        if not check:
+            return rc, stdout, stderr
+
+        if rc != 0:
+            print(
+                f"SSH command {cmd} exited with non zero error code: {rc}\n"
+                f"stdout: {stdout}\n"
+                f"stderr: {stderr}\n"
+            )
+
             if self._on_error:
-                self._on_error(exc)
+                self._on_error()
+
+            assert False
 
-            raise
+        return rc, stdout, stderr
 
     # pylint:disable=invalid-name
     def Popen(

tests/integration_tests/functional/test_balloon.py

@@ -41,23 +41,11 @@ def get_rss_from_pmap():
 
 def lower_ssh_oom_chance(ssh_connection):
     """Lure OOM away from ssh process"""
-    logger = logging.getLogger("lower_ssh_oom_chance")
-
     cmd = "cat /run/sshd.pid"
-    exit_code, stdout, stderr = ssh_connection.run(cmd)
-    # add something to the logs for troubleshooting
-    if exit_code != 0:
-        logger.error("while running: %s", cmd)
-        logger.error("stdout: %s", stdout)
-        logger.error("stderr: %s", stderr)
-
+    _, stdout, _ = ssh_connection.run(cmd)
     for pid in stdout.split(" "):
         cmd = f"choom -n -1000 -p {pid}"
-        exit_code, stdout, stderr = ssh_connection.run(cmd)
-        if exit_code != 0:
-            logger.error("while running: %s", cmd)
-            logger.error("stdout: %s", stdout)
-            logger.error("stderr: %s", stderr)
+        ssh_connection.run(cmd)
 
 
 def make_guest_dirty_memory(ssh_connection, amount_mib=32):
@@ -68,12 +56,7 @@ def make_guest_dirty_memory(ssh_connection, amount_mib=32):
 
     cmd = f"/usr/local/bin/fillmem {amount_mib}"
     try:
-        exit_code, stdout, stderr = ssh_connection.run(cmd, timeout=1.0)
-        # add something to the logs for troubleshooting
-        if exit_code != 0:
-            logger.error("while running: %s", cmd)
-            logger.error("stdout: %s", stdout)
-            logger.error("stderr: %s", stderr)
+        ssh_connection.run(cmd, timeout=1.0)
     except TimeoutExpired:
         # It's ok if this expires. Sometimes the SSH connection
         # gets killed by the OOM killer *after* the fillmem program
@@ -558,4 +541,4 @@ def test_memory_scrub(microvm_factory, guest_kernel, rootfs):
     # Wait for the deflate to complete.
     _ = get_stable_rss_mem_by_pid(firecracker_pid)
 
-    microvm.ssh.check_output("/usr/local/bin/readmem {} {}".format(60, 1))
+    microvm.ssh.run("/usr/local/bin/readmem {} {}".format(60, 1))

tests/integration_tests/functional/test_cpu_features_aarch64.py

@@ -51,7 +51,7 @@ def _check_cpu_features_arm(test_microvm, guest_kv, template_name=None):
         case CpuModel.ARM_NEOVERSE_V1, _, None:
             expected_cpu_features = DEFAULT_G3_FEATURES_5_10
 
-    _, stdout, _ = test_microvm.ssh.check_output(CPU_FEATURES_CMD)
+    _, stdout, _ = test_microvm.ssh.run(CPU_FEATURES_CMD)
     flags = set(stdout.strip().split(" "))
     assert flags == expected_cpu_features
 
@@ -77,7 +77,7 @@ def test_host_vs_guest_cpu_features_aarch64(uvm_nano):
     vm.add_net_iface()
     vm.start()
     host_feats = set(utils.check_output(CPU_FEATURES_CMD).stdout.strip().split(" "))
-    guest_feats = set(vm.ssh.check_output(CPU_FEATURES_CMD).stdout.strip().split(" "))
+    guest_feats = set(vm.ssh.run(CPU_FEATURES_CMD).stdout.strip().split(" "))
 
     cpu_model = cpuid_utils.get_cpu_model_name()
     match cpu_model:

tests/integration_tests/functional/test_cpu_features_x86_64.py

@@ -215,7 +215,7 @@ def test_host_vs_guest_cpu_features_x86_64(uvm_nano):
     vm.add_net_iface()
     vm.start()
     host_feats = set(utils.check_output(CPU_FEATURES_CMD).stdout.strip().split(" "))
-    guest_feats = set(vm.ssh.check_output(CPU_FEATURES_CMD).stdout.strip().split(" "))
+    guest_feats = set(vm.ssh.run(CPU_FEATURES_CMD).stdout.strip().split(" "))
 
     cpu_model = cpuid_utils.get_cpu_codename()
     match cpu_model:

tests/integration_tests/functional/test_drive_vhost_user.py

@@ -15,8 +15,7 @@ def _check_block_size(ssh_connection, dev_path, size):
     """
     Checks the size of the block device.
     """
-    _, stdout, stderr = ssh_connection.run("blockdev --getsize64 {}".format(dev_path))
-    assert stderr == ""
+    _, stdout, _ = ssh_connection.run("blockdev --getsize64 {}".format(dev_path))
     assert stdout.strip() == str(size)
 
 
@@ -297,7 +296,7 @@ def test_config_change(microvm_factory, guest_kernel, rootfs):
     _check_block_size(vm.ssh, "/dev/vdb", orig_size * 1024 * 1024)
 
     # Check that we can create a filesystem and mount it
-    vm.ssh.check_output(mkfs_mount_cmd)
+    vm.ssh.run(mkfs_mount_cmd)
 
     for new_size in new_sizes:
         # Instruct the backend to resize the device.
@@ -312,4 +311,4 @@ def test_config_change(microvm_factory, guest_kernel, rootfs):
         _check_block_size(vm.ssh, "/dev/vdb", new_size * 1024 * 1024)
 
         # Check that we can create a filesystem and mount it
-        vm.ssh.check_output(mkfs_mount_cmd)
+        vm.ssh.run(mkfs_mount_cmd)

tests/integration_tests/functional/test_drive_virtio.py

@@ -283,7 +283,7 @@ def test_patch_drive(uvm_plain_any, io_engine):
     # of the device, in bytes.
     blksize_cmd = "LSBLK_DEBUG=all lsblk -b /dev/vdb --output SIZE"
     size_bytes_str = "536870912"  # = 512 MiB
-    _, stdout, _ = test_microvm.ssh.check_output(blksize_cmd)
+    _, stdout, _ = test_microvm.ssh.run(blksize_cmd)
     lines = stdout.split("\n")
     # skip "SIZE"
     assert lines[1].strip() == size_bytes_str
@@ -354,14 +354,12 @@ def test_flush(uvm_plain_rw, io_engine):
 
 
 def _check_block_size(ssh_connection, dev_path, size):
-    _, stdout, stderr = ssh_connection.run("blockdev --getsize64 {}".format(dev_path))
-    assert stderr == ""
+    _, stdout, _ = ssh_connection.run("blockdev --getsize64 {}".format(dev_path))
     assert stdout.strip() == str(size)
 
 
 def _check_file_size(ssh_connection, dev_path, size):
-    _, stdout, stderr = ssh_connection.run("stat --format=%s {}".format(dev_path))
-    assert stderr == ""
+    _, stdout, _ = ssh_connection.run("stat --format=%s {}".format(dev_path))
     assert stdout.strip() == str(size)
 
 
@@ -379,7 +377,5 @@ def _check_drives(test_microvm, assert_dict, keys_array):
 
 
 def _check_mount(ssh_connection, dev_path):
-    _, _, stderr = ssh_connection.run(f"mount {dev_path} /tmp", timeout=30.0)
-    assert stderr == ""
-    _, _, stderr = ssh_connection.run("umount /tmp", timeout=30.0)
-    assert stderr == ""
+    ssh_connection.run(f"mount {dev_path} /tmp", timeout=30.0)
+    ssh_connection.run("umount /tmp", timeout=30.0)