3proxy/tests/cases/reload.py
Vladimir Dubrovin 21ba6e9d0c Reload the configuration atomically
A reload reads the file into a new configuration, allocated from the
defaults, and prepares its services without starting them. Only if the
whole file is read without an error does it come into force: the old
configuration's services stop accepting, the new ones start. On an
error everything prepared is dropped and the old configuration, which
never stopped, stays in force.

Each service holds its configuration until it is freed, after its last
client; a configuration no longer in force is released when the last
holder lets go (confhold/confput). freeconf() now only releases what a
configuration holds, touching nothing global.

What runs without a service of its own reads the configuration in
force, runconf, rather than conf, which a reload points at the one
being read: the resolver, which holds it while it waits (a client
resolves with its own service's), the log, the thread pool, the admin
pages, which walk the services under config_mutex, and the signal
handlers.

The users table, havelog and the resolver choice are per configuration
now, the resolver applied when the configuration comes into force, so
a refused file leaves the users as they were. The log file, the
configuration file name, the string table, daemon mode and the filter
list belong to the process and are handed from one configuration to
the next. Settings left out of a new file go back to their defaults,
where before they kept the old values.

The gap between the old and the new listening sockets remains: a
connection queued on an old service's socket when it closes is reset.
Handing the socket over is the next phase.

A test case reloads a broken file (nothing changes, alice still gets
in, bob from the refused file does not) and a good one (bob only, the
service on its new port, the old one closed, a download through the
old service finishing whole). 40 reloads, a third refused, under load
show no memory errors under AddressSanitizer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 20:15:14 +03:00

104 lines
3.8 KiB
Python

"""Reloading the configuration: a file with an error leaves the configuration
in force, services and users, as it was; a good one replaces it, while the
connections of the old services finish undisturbed."""
import threading
import time
def config(origin, proxy, adm, user, extra=""):
return f"""
log
auth iponly
allow *
http data * /data
httpsrv -p{origin}
flush
users {user}
auth strong
allow *
proxy -p{proxy}
flush
auth iponly
allow *
admin -p{adm}
{extra}
"""
def run(t):
origin = t.free_port()
first = t.free_port()
second = t.free_port()
adm = t.free_port()
server = t.start("reload", config(origin, first, adm, "alice:CL:one"),
ports=[origin, first, adm])
url = f"http://127.0.0.1:{origin}/data?size=1000"
proxy = f"127.0.0.1:{first}"
def as_user(user, password, through=proxy):
return t.http(url, proxy=through, proxy_auth=(user, password)).status
t.eq(200, as_user("alice", "one"), "the first configuration lets alice in")
def reload_and_wait(expect_failure=False):
before = server.output()
accepting = before.count("Accepting connections")
t.http(f"http://127.0.0.1:{adm}/R")
deadline = time.time() + 10
while time.time() < deadline:
out = server.output()
if expect_failure and "Unknown command" in out[len(before):]:
return out
if not expect_failure and out.count("Accepting connections") >= accepting + 3:
return out
time.sleep(0.05)
return server.output()
# --- a file with an error: nothing changes --------------------------
t.write_config("reload", config(origin, first, adm, "bob:CL:two", "nosuchcommand"))
out = reload_and_wait(expect_failure=True)
t.contains(out, "Unknown command", "the broken configuration is refused")
time.sleep(1.5) # what a switch would take: the old services stop within a second
t.eq(200, as_user("alice", "one"), "after a failed reload alice still gets in")
t.ne(200, as_user("bob", "two"), "and bob, from the refused file, does not")
# --- a good file: the new configuration takes over -------------------
slow = {}
def download():
# about two seconds: 2 MB in 64 KB blocks, 60 ms apart
r = t.http(f"http://127.0.0.1:{origin}/data?size=2000000&block=65536&delay=60",
proxy=proxy, proxy_auth=("alice", "one"))
slow["status"], slow["length"] = r.status, r.length
th = threading.Thread(target=download)
th.start()
time.sleep(0.3)
t.write_config("reload", config(origin, second, adm, "bob:CL:two"))
reload_and_wait()
# The old services stop accepting within a second; until then both the
# old and the new origin listen, and a connection the old one has queued
# is reset when it closes (a listener is not handed over yet).
time.sleep(1.5)
t.ok("the new services start") if t.wait_port(second) else \
t.fail("the new services start", f"port {second} open", "closed")
moved = f"127.0.0.1:{second}"
t.eq(200, as_user("bob", "two", moved), "the new configuration lets bob in")
t.ne(200, as_user("alice", "one", moved), "and not alice, who is gone from it")
# the old service stops accepting, its port closes
deadline = time.time() + 5
while time.time() < deadline and t.wait_port(first, timeout=0.2):
time.sleep(0.2)
t.ok("the old service stops listening") if not t.wait_port(first, timeout=0.2) else \
t.fail("the old service stops listening", f"port {first} closed", "open")
th.join(20)
t.eq(200, slow.get("status"), "a download through the old service finishes")
t.eq(2000000, slow.get("length"), "with the whole answer")