fix: improve Dial reliability for slow exporters

docs/source/getting-started/configuration/files.md

@@ -46,6 +46,10 @@ token: "******************" # An authentication token
 drivers:
   allow: ["jumpstarter_drivers_*", "vendorpackage.*"] # Driver packages the client can dynamically load
   unsafe: false # Allow any driver package to load dynamically
+leases:
+  acquisition_timeout: 7200 # Timeout in seconds for lease acquisition (default: 7200)
+  dial_timeout: 60 # Time limit for the exporter to become ready after lease is acquired (default: 60)
+  retry_timeout: 300 # Time limit for re-attempting a lease when the exporter becomes unreachable (default: 300, 0 to disable)
 ```
 
 **Environment Variables**:
@@ -59,6 +63,16 @@ drivers:
 - `JMP_TOKEN` - Auth token (overrides config file)
 - `JMP_DRIVERS_ALLOW` - Comma-separated list of allowed driver namespaces
 - `JUMPSTARTER_FORCE_SYSTEM_CERTS` - Set to `1` to force system CA certificates
+- `JMP_RETRY_TIMEOUT` - Retry timeout in seconds for unreachable exporters (overrides config, default: 300)
+- `JMP_DIAL_TIMEOUT` - Dial timeout in seconds for slow exporters (overrides config, default: 60)
+- `JMP_OIDC_CALLBACK_PORT` - Local port for the OIDC callback during `jmp login` (useful for SSH tunneling; default: OS-assigned)
+- `JMP_GRPC_PASSPHRASE` - Shared passphrase for authenticating against passphrase-protected exporters
+
+**Shell Session Variables** (automatically set by `jmp shell`):
+
+- `JMP_LEASE` - Active lease name (enables reconnection via `JMP_LEASE=<name> jmp shell`)
+- `JMP_EXPORTER` - Name of the connected exporter
+- `JMP_EXPORTER_LABELS` - Connected exporter's labels as comma-separated `key=value` pairs
 
 **CLI Commands**:
 ```{code-block}  console
@@ -133,6 +147,7 @@ boundaries. See [{term}`Hook`s](../../introduction/hooks.md) for full details on
 - `JMP_TOKEN` - Auth token (overrides config file)
 - `JMP_NAMESPACE` - Namespace in the {term}`controller`
 - `JMP_NAME` - {term}`Exporter` name
+- `JMP_DISABLE_COMPRESSION` - Set to `1` to disable stream compression (gzip, xz, bz2, zstd) for driver data transfers
 
 **CLI Commands**:
 ```{code-block}  console

docs/source/getting-started/configuration/loading-order.md

@@ -17,9 +17,9 @@ precedence (highest to lowest):
 
 For client operations, Jumpstarter processes configurations in this order:
 
-1. **Command-line options** such as `--endpoint` or `--client-config`
-2. **Environment variables** such as `JMP_ENDPOINT`, `JMP_TOKEN`, or
-   `JMP_CLIENT_CONFIG`
+1. **Command-line options** such as `--endpoint`, `--client-config`, `--retry-timeout`, or `--dial-timeout`
+2. **Environment variables** such as `JMP_ENDPOINT`, `JMP_TOKEN`,
+   `JMP_CLIENT_CONFIG`, `JMP_RETRY_TIMEOUT`, or `JMP_DIAL_TIMEOUT`
 3. **Current client** defined in `${HOME}/.config/jumpstarter/config.yaml`
 4. **Specific client file** in `${HOME}/.config/jumpstarter/clients/<n>.yaml`
 

python/packages/jumpstarter-cli/jumpstarter_cli/common.py

@@ -159,6 +159,20 @@ def convert(self, value, param, ctx):
     ),
 )
 
+DIAL_TIMEOUT = DurationParamType(minimum=timedelta(seconds=5))
+
+opt_dial_timeout = partial(
+    click.option,
+    "--dial-timeout",
+    "dial_timeout",
+    type=DIAL_TIMEOUT,
+    default=None,
+    help=(
+        "Override dial timeout for slow exporters (e.g., '60s', '2m', "
+        "'90s'). Env: JMP_DIAL_TIMEOUT. Default: 60s."
+    ),
+)
+
 opt_begin_time = click.option(
     "--begin-time",
     "begin_time",

python/packages/jumpstarter-cli/jumpstarter_cli/shell.py

@@ -24,7 +24,14 @@
 )
 from jumpstarter_cli_common.signal import signal_handler
 
-from .common import opt_acquisition_timeout, opt_duration_partial, opt_exporter_name, opt_retry_timeout, opt_selector
+from .common import (
+    opt_acquisition_timeout,
+    opt_dial_timeout,
+    opt_duration_partial,
+    opt_exporter_name,
+    opt_retry_timeout,
+    opt_selector,
+)
 from .login import relogin_client
 from jumpstarter.client import DirectLease
 from jumpstarter.client.client import client_from_path, fetch_motd
@@ -473,7 +480,7 @@ async def _run_shell_with_lease_async(lease, exporter_logs, config, command, can
 
 async def _shell_with_signal_handling(  # noqa: C901
     config, selector, exporter_name, lease_name, duration, exporter_logs, command, acquisition_timeout,
-    retry_timeout=None,
+    retry_timeout=None, dial_timeout=None,
 ):
     """Handle lease acquisition and shell execution with signal handling."""
     exit_code = 0
@@ -500,7 +507,7 @@ async def _shell_with_signal_handling(  # noqa: C901
                     while True:
                         async with config.lease_async(
                             selector, exporter_name, lease_name, duration, portal, acquisition_timeout,
-                            retry_timeout=retry_timeout,
+                            retry_timeout=retry_timeout, dial_timeout=dial_timeout,
                         ) as lease:
                             lease_used = lease
 
@@ -521,6 +528,11 @@ async def _shell_with_signal_handling(  # noqa: C901
                             if unreachable is not None:
                                 if lease.lease_ended:
                                     break  # lease expired naturally — exit cleanly
+                                if lease.lease_transferred:
+                                    raise ExporterOfflineError(
+                                        "Lease has been transferred to another client. "
+                                        "Session is no longer valid."
+                                    ) from unreachable
                                 if connect_deadline is None:
                                     connect_deadline = time.monotonic() + lease.retry_timeout
                                 if time.monotonic() >= connect_deadline:
@@ -689,6 +701,7 @@ async def _shell_direct_async(
 @click.option("--exporter-logs", is_flag=True, help="Enable exporter log streaming")
 @opt_acquisition_timeout()
 @opt_retry_timeout()
+@opt_dial_timeout()
 # direct connection (no controller)
 @click.option(
     "--tls-grpc",
@@ -720,6 +733,7 @@ def shell(
     exporter_logs,
     acquisition_timeout,
     retry_timeout,
+    dial_timeout,
     tls_grpc_address,
     tls_grpc_insecure,
     passphrase,
@@ -769,6 +783,7 @@ def shell(
                 command,
                 acquisition_timeout,
                 retry_timeout,
+                dial_timeout,
             )
             sys.exit(exit_code)
 

python/packages/jumpstarter-cli/jumpstarter_cli/shell_test.py

@@ -68,7 +68,7 @@ def __init__(self):
     @asynccontextmanager
     async def lease_async(
         self, selector, exporter_name, lease_name, duration, portal,
-        acquisition_timeout, retry_timeout=None,
+        acquisition_timeout, retry_timeout=None, dial_timeout=None,
     ):
         self.captured = (selector, exporter_name, lease_name, duration, acquisition_timeout)
         m = Mock()
@@ -113,7 +113,7 @@ async def test_shell_warns_when_expired_token_prevents_cleanup_on_normal_exit():
     @asynccontextmanager
     async def lease_async(
         selector, exporter_name, lease_name, duration, portal,
-        acquisition_timeout, retry_timeout=None,
+        acquisition_timeout, retry_timeout=None, dial_timeout=None,
     ):
         yield lease
 
@@ -162,6 +162,7 @@ def test_shell_requires_selector_or_name_when_no_leases():
             exporter_logs=False,
             acquisition_timeout=None,
             retry_timeout=None,
+            dial_timeout=None,
             tls_grpc_address=None,
             tls_grpc_insecure=False,
             passphrase=None,
@@ -183,6 +184,7 @@ def test_shell_allows_existing_lease_name_without_selector_or_name():
             exporter_logs=False,
             acquisition_timeout=None,
             retry_timeout=None,
+            dial_timeout=None,
             tls_grpc_address=None,
             tls_grpc_insecure=False,
             passphrase=None,
@@ -208,6 +210,7 @@ def test_shell_auto_connects_single_lease():
             exporter_logs=False,
             acquisition_timeout=None,
             retry_timeout=None,
+            dial_timeout=None,
             tls_grpc_address=None,
             tls_grpc_insecure=False,
             passphrase=None,
@@ -236,6 +239,7 @@ def test_shell_no_leases_shows_guidance():
             exporter_logs=False,
             acquisition_timeout=None,
             retry_timeout=None,
+            dial_timeout=None,
             tls_grpc_address=None,
             tls_grpc_insecure=False,
             passphrase=None,
@@ -277,6 +281,7 @@ def test_shell_multi_lease_no_tty_error():
             exporter_logs=False,
             acquisition_timeout=None,
             retry_timeout=None,
+            dial_timeout=None,
             tls_grpc_address=None,
             tls_grpc_insecure=False,
             passphrase=None,
@@ -313,6 +318,7 @@ def test_shell_no_own_leases_among_others():
             exporter_logs=False,
             acquisition_timeout=None,
             retry_timeout=None,
+            dial_timeout=None,
             tls_grpc_address=None,
             tls_grpc_insecure=False,
             passphrase=None,
@@ -335,6 +341,7 @@ def test_shell_allows_env_lease_without_selector_or_name():
             exporter_logs=False,
             acquisition_timeout=None,
             retry_timeout=None,
+            dial_timeout=None,
             tls_grpc_address=None,
             tls_grpc_insecure=False,
             passphrase=None,
@@ -1053,9 +1060,9 @@ def _make_config_with_lease(self, lease):
 
         @asynccontextmanager
         async def lease_async(
-        selector, exporter_name, lease_name, duration, portal,
-        acquisition_timeout, retry_timeout=None,
-    ):
+            selector, exporter_name, lease_name, duration, portal,
+            acquisition_timeout, retry_timeout=None, dial_timeout=None,
+        ):
             yield lease
 
         config.lease_async = lease_async
@@ -1136,9 +1143,9 @@ async def test_retries_then_raises_on_timeout(self):
 
         @asynccontextmanager
         async def lease_async(
-        selector, exporter_name, lease_name, duration, portal,
-        acquisition_timeout, retry_timeout=None,
-    ):
+            selector, exporter_name, lease_name, duration, portal,
+            acquisition_timeout, retry_timeout=None, dial_timeout=None,
+        ):
             yield lease
 
         config.lease_async = lease_async
@@ -1180,9 +1187,9 @@ async def test_retries_when_wrapped_in_exception_group(self):
 
         @asynccontextmanager
         async def lease_async(
-        selector, exporter_name, lease_name, duration, portal,
-        acquisition_timeout, retry_timeout=None,
-    ):
+            selector, exporter_name, lease_name, duration, portal,
+            acquisition_timeout, retry_timeout=None, dial_timeout=None,
+        ):
             yield lease
 
         config.lease_async = lease_async
@@ -1222,9 +1229,9 @@ async def test_retry_succeeds_before_timeout(self):
 
         @asynccontextmanager
         async def lease_async(
-        selector, exporter_name, lease_name, duration, portal,
-        acquisition_timeout, retry_timeout=None,
-    ):
+            selector, exporter_name, lease_name, duration, portal,
+            acquisition_timeout, retry_timeout=None, dial_timeout=None,
+        ):
             yield lease
 
         config.lease_async = lease_async
@@ -1266,7 +1273,7 @@ async def test_exits_cleanly_when_lease_ended_during_connection(self):
         @asynccontextmanager
         async def lease_async(
             selector, exporter_name, lease_name, duration, portal,
-            acquisition_timeout, retry_timeout=None,
+            acquisition_timeout, retry_timeout=None, dial_timeout=None,
         ):
             yield lease
 
@@ -1303,7 +1310,7 @@ async def test_retries_normally_when_lease_not_ended(self):
         @asynccontextmanager
         async def lease_async(
             selector, exporter_name, lease_name, duration, portal,
-            acquisition_timeout, retry_timeout=None,
+            acquisition_timeout, retry_timeout=None, dial_timeout=None,
         ):
             yield lease