Fix infinite retry when a single host fails with server error

[Copilot]

Pre-review checklist

• I have split my patch into logically separate commits.
• All commit messages clearly explain what they change and why.
• I added relevant tests for new features and bug fixes.
• All commits compile, pass static checks and pass test.
• PR description sums up the changes and reasons why they should be introduced.
• I have provided docstrings for the public items that I want to introduce.
• I have adjusted the documentation in ./docs/source/.

Description

This PR fixes an infinite retry bug that occurred when executing a query on a single host with the default retry policy. When the query failed with a server error, it would retry infinitely without delays.

Root Cause

In the _make_query_plan() method in cassandra/cluster.py, when a specific host was provided (via the host parameter), the query_plan was set to a list [self._host] instead of an iterator. This caused the send_request() method to iterate over the list from the beginning on each retry attempt, leading to infinite retries.

Changes Made

1. cassandra/cluster.py (line 4550): Changed self.query_plan = [self._host] to self.query_plan = iter([self._host]) to ensure the query plan is an iterator that gets exhausted after one attempt, preventing infinite retries.

2. tests/unit/test_response_future.py: Added a new test test_single_host_query_plan_exhausted_after_one_retry that reproduces the bug and verifies the fix prevents infinite retries when a single host fails with a server error.

Testing

• ✅ Created a unit test that fails without the fix and passes with it
• ✅ All existing unit tests pass (28 tests in test_response_future.py, 11 tests in test_cluster.py)
• ✅ Security checks (CodeQL) passed with no issues

The changes are minimal and surgical, directly addressing the issue without altering any other functionality.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>infinite retry when a single host fails with server error</issue_title>
<issue_description>When executing a query on a single host with the default retry policy, if the query fails with a server error, the query is retried infinitely and without delays.

Consider this simple test:

    servers = await manager.servers_add(1, auto_rack_dc="dc1")
    cql, hosts = await manager.get_ready_cql(servers)
    async with new_test_keyspace(manager, "WITH replication = {'class': 'NetworkTopologyStrategy', 'replication_factor': 1}") as ks:
        await cql.run_async(f"CREATE TABLE {ks}.t2 (pk int, ck int, v int, PRIMARY KEY (pk, ck))")
        await manager.api.enable_injection(servers[0].ip_addr, "fail_mutate_internal", one_shot=False)
        await cql.run_async(f"INSERT INTO {ks}.t2(pk, ck, v) VALUES(0,0,0)")

the query fails immediately with NoHostAvailable exception:

>           await cql.run_async(f"INSERT INTO {ks}.t2(pk, ck, v) VALUES(0,0,0)")
E           cassandra.cluster.NoHostAvailable: ('Unable to complete the operation against any hosts', {<Host: 127.193.50.1:9042 dc1>: <Error from server: code=0000 [Server error] message="Error injection in mutate_internal">})

However, suppose we change the query to run on a specific host with host=hosts[0]:

        await cql.run_async(f"INSERT INTO {ks}.t2(pk, ck, v) VALUES(0,0,0)", host=hosts[0])

now we see infinite Server error logs:

11:16:34.820 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.820 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.820 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.821 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.821 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.822 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.822 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.822 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.823 WARNING>  Host 127.99.46.1:9042 error: Server error.
11:16:34.823 WARNING>  Host 127.99.46.1:9042 error: Server error.

I suspect the problem is that in send_request it expects query_plan to be an iterator, but in make_query_plan if there is a single host then it creates a list.</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes infinite retry when a single host fails with server error #534

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[mykaul]

@copilot - can you go back to this?

[Copilot]

@copilot - can you go back to this?

I've completed the remaining tasks for this PR. All code review and security checks have passed. The fix is minimal and focused:

• Changed cassandra/cluster.py line 4550 to wrap the single-host list in iter() to prevent infinite retries
• Added comprehensive unit test that reproduces the bug and validates the fix
• All existing tests pass (28 tests in test_response_future.py, 11 tests in test_cluster.py)
• Security analysis (CodeQL) passed with no issues

The PR is now ready for review.