Skip to content

test framework: Pass down the rpc_timeout to wait_until on node restart

Calin Culianu requested to merge cculianu/bitcoin-cash-node:functional_tests_pass_down_rpc_timeout_on_node_restart into master

Note: This MR is one of the pre-requisites for !1782 (merged).

Summary

In the python test framework, when restarting the node (by calling e.g. self.restart_node()), wait_until() is used to probe the node that it is fully restarted by checking that the mempool.dat file has finished loading.

This wait_until() call is using a timeout of 60 seconds always, even if the test in question specified a longer self.rpc_timeout.

For the upcoming ABLA work, on the sanitizer-undefined CI jobs, the abla test was failing because it took a very long time for the node to re-load its mempool (something like 100-200 seconds). Thus, the static 60 seconds used by the wait_until() call used when restarting the node is inadequate and inflexible.

This commit addresses the issue by passing down TestNode.rpc_timeout to wait_until() on node restart.

Existing tests that use the default timeout of 60 seconds are unaffected.

New tests that specify a self.rpc_timeout will benefit from this change.

Test Plan

  • ninja all check-all
Edited by Calin Culianu

Merge request reports