The ab(1) command we use for testing is limited to 20000 open
connections under Debian jessie; a perfectly reasonable limit
to avoid port exhaustion. I never noticed this limit before,
but systemd under Jessie seems to have upped the default
RLIMIT_NOFILE to 65536(!), causing ab to error out.
We don't even need 10K connections for testing,
we just need to hit *some* limit before we start expiring.
So lower the RLIMIT_NOFILE back to 1024 in the forked server
process so we can test more quickly without running out of
ports or memory, since exhausting the 65536 RLIMIT_NOFILE
limit is not going to happen with a single TCP address.
---
test/test_client_expire.rb | 13 +++++++++++--
1 file changed, 11 insertions(+), 2 deletions(-)
diff --git a/test/test_client_expire.rb b/test/test_client_expire.rb
index b2f932b..b6dec1a 100644
--- a/test/test_client_expire.rb
+++ b/test/test_client_expire.rb
@@ -99,6 +99,14 @@ def test_client_expire_desperate
end
stderr_path err.path
end
+
+ # 1024 is common on old systems, but nowadays Jessie seems to be 65536
+ nr = Process.getrlimit(:NOFILE)[0]
+ if nr >= 1024
+ nr = 1024
+ do_rlimit = true
+ end
+
pid = mkserver(cfg) do
keep = { $stderr => true, $stdout => true, $stdin => true, @srv => true }
ObjectSpace.each_object(IO) do |obj|
@@ -108,7 +116,9 @@ def test_client_expire_desperate
rescue IOError # could be uninitialized
end
end
+ Process.setrlimit(:NOFILE, nr) if do_rlimit
end
+
f = get_tcp_client(host, port)
f.write "G"
s = get_tcp_client(host, port)
@@ -119,10 +129,9 @@ def test_client_expire_desperate
assert_match(%r{keep-alive}, str)
sleep 1
- # ignore errors, just beat the crap out of the process
- nr = Process.getrlimit(:NOFILE)[0] # 1024 is common
assert_operator nr, :>, 666, "increase RLIM_NOFILE (ulimit -n)"
nr -= 50
+ # ignore errors, just beat the crap out of the process
opts = { out: "/dev/null", err: "/dev/null", close_others: true }
begin
pids = 2.times.map do
--
unsubscribe: [email protected]
archive: https://yhbt.net/yahns-public/
