pytest: add failing downgrade test.
What the developer wrote
Authored by Rusty Russell
pytest: add failing downgrade test.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
What changed, and why it matters
This commit only adds a new test to the project's test suite. It checks that the database downgrade tool correctly refuses to downgrade when a newer feature (a node bias in the 'askrene' routing layer) is present. There is no change to production code, no fix for a vulnerability, and no security-relevant behavior change in the software itself.
No security action required. Treat as normal test-suite maintenance. If reviewing a larger patch series, evaluate the accompanying production-code changes separately.
Security signals we found
No strong security signals were identified.
Evidence from the diff
The commit refactors an existing pytest file (tests/test_downgrade.py) by extracting the command-line construction for tools/lightning-downgrade into a helper function, then adds a new test, test_downgrade_fail. The new test creates a node bias via askrene_bias_node, stops the node, runs the downgrade tool, and asserts that it exits with ERROR_DBFAIL and prints a message indicating the node bias is unsupported in v25.09. This is purely test infrastructure and regression coverage; it does not modify the downgrade tool or any runtime code.
Changed components
tests/test_downgrade.pyInspect captured patch +40 / −19
diff --git a/tests/test_downgrade.py b/tests/test_downgrade.py
index a9e07f2..e460f3d 100644
--- a/tests/test_downgrade.py
+++ b/tests/test_downgrade.py
@@ -7,6 +7,29 @@ from utils import (
import os
import subprocess
+# From the binary:
+# ERROR_DBVERSION = 1
+ERROR_DBFAIL = 2
+ERROR_USAGE = 3
+# ERROR_INTERNAL = 99
+
+
+def downgrade_cmdline(node):
+ # lightning-downgrade understands a subset of the options
+ # to lightningd.
+ downgrade_opts = []
+ for o in node.daemon.opts:
+ if o in ('network', 'lightning-dir', 'conf', 'rpc-file', 'wallet'):
+ if node.daemon.opts[o] is None:
+ downgrade_opts.append(f"--{o}")
+ else:
+ downgrade_opts.append(f"--{o}={node.daemon.opts[o]}")
+
+ cmd_line = ["tools/lightning-downgrade"] + downgrade_opts
+ if os.getenv("VALGRIND") == "1":
+ cmd_line = ['valgrind', '-q', '--error-exitcode=7'] + cmd_line
+ return cmd_line
+
def test_downgrade(node_factory, executor):
l1, l2 = node_factory.line_graph(2, opts={'may_reconnect': True}, wait_for_announce=True)
@@ -22,25 +45,7 @@ def test_downgrade(node_factory, executor):
old_inv = l2.rpc.invoice(1000, 'test_downgrade1', 'test_downgrade')
l1.rpc.xpay(old_inv['bolt11'])
- # From the binary:
- # ERROR_DBVERSION = 1
- # ERROR_DBFAIL = 2
- ERROR_USAGE = 3
- # ERROR_INTERNAL = 99
-
- # lightning-downgrade understands a subset of the options
- # to lightningd.
- downgrade_opts = []
- for o in l1.daemon.opts:
- if o in ('network', 'lightning-dir', 'conf', 'rpc-file', 'wallet'):
- if l1.daemon.opts[o] is None:
- downgrade_opts.append(f"--{o}")
- else:
- downgrade_opts.append(f"--{o}={l1.daemon.opts[o]}")
-
- cmd_line = ["tools/lightning-downgrade"] + downgrade_opts
- if os.getenv("VALGRIND") == "1":
- cmd_line = ['valgrind', '-q', '--error-exitcode=7'] + cmd_line
+ cmd_line = downgrade_cmdline(l1)
# No downgrade on live nodes!
retcode = subprocess.call(cmd_line, timeout=TIMEOUT)
@@ -89,3 +94,19 @@ def test_downgrade(node_factory, executor):
bias = only_one(only_one(l1.rpc.askrene_listlayers('xpay')['layers'])['biases'])
assert bias['short_channel_id_dir'] == bias_scidd
assert bias['bias'] == 1
+
+
+def test_downgrade_fail(node_factory, executor):
+ """If we have created as node bias, we cannot downgrade"""
+ l1, l2 = node_factory.line_graph(2, opts={'may_reconnect': True}, wait_for_announce=True)
+
+ l1.rpc.askrene_bias_node('xpay', l2.info['id'], 'in', 1)
+ cmd_line = downgrade_cmdline(l1)
+
+ l1.stop()
+
+ p = subprocess.Popen(cmd_line, stdout=subprocess.DEVNULL,
+ stderr=subprocess.PIPE)
+ _, err = p.communicate(timeout=TIMEOUT)
+ assert p.returncode == ERROR_DBFAIL
+ assert 'Askrene has a node bias, which is not supported in v25.09' in err.decode('utf-8')
Why this scored 12/100
Community notes
Notes can correct, qualify, or add evidence to the AI analysis. Every note shown here has been validated by a human moderator.
The AI analysis stands alone for now. Submit a note if you can add evidence or important context.