Dualstack interop testing enablement #11231
Conversation
| @@ -268,7 +267,7 @@ void setUp() { | |||
| tester.setUp(); | |||
| } | |||
|
|
|||
| private synchronized void tearDown() { | |||
| synchronized void tearDown() { | |||
There was a problem hiding this comment.
Why was this file changed? I don't see any usages of these changes.
| InetAddress[] addresses = InetAddress.getAllByName(InetAddress.getLocalHost().getHostName()); | ||
| for (InetAddress address : addresses) { | ||
| if (address.getAddress().length == 4) { | ||
| return new java.net.InetSocketAddress(address.getHostAddress(), port); |
There was a problem hiding this comment.
Pass address, like you did for IPv6. This will do a DNS lookup and can leave you with an IPV6 address.
| @@ -130,11 +138,14 @@ void parseArgs(String[] args) { | |||
| + "\n --local_handshaker_port=PORT" | |||
| + "\n Use local ALTS handshaker service on the specified port " | |||
| + "\n for testing. Only effective when --use_alts=true." | |||
| + "\n --address_type=IPV4|IPV6|IPV4_IPV6" | |||
There was a problem hiding this comment.
Will there be an update to https://github.com/grpc/grpc/blob/master/doc/interop-test-descriptions.md ?
| @@ -162,11 +174,15 @@ private void parseArgs(String[] args) { | |||
| + s.secureMode | |||
| + "\n --server_id=STRING server ID for response." | |||
| + "\n Default: " | |||
| + s.serverId); | |||
| + s.serverId | |||
| + "\n --address_type=STRING type of IP address to bind to (IPV4|IPV6|IPV4_IPV6)." | |||
There was a problem hiding this comment.
Will there be an update to https://github.com/grpc/grpc/blob/master/doc/xds-test-descriptions.md ?
| @@ -108,58 +111,112 @@ public void allCommandlineSwitchesAreSupported() { | |||
| assertEquals(60, client.metricsLogRateSecs()); | |||
| } | |||
|
|
|||
| @Test | |||
| public void gaugesShouldBeExported_ipv4() throws Exception { | |||
| checkGaugesShouldBeExported(IPV4); | |||
There was a problem hiding this comment.
Is there any document describing why we are messing with the stress client?
There was a problem hiding this comment.
Because it is the only test that actually talks to a server over IP.
There was a problem hiding this comment.
So make a new test? I don't follow. It doesn't even need a client; you could just open a TCP connection to see if the port is open, if you wanted.
What clued me in to this is the switch statement. We really want to avoid logic in tests.
larry-safran
force-pushed
the
dualstack_server_flags
branch
from
de679ad
to
4a97c3c
Compare
…tional flag "address_type"
larry-safran
force-pushed
the
dualstack_server_flags
branch
from
a328727
to
96388a2
Compare
Fix flag string for enabling dual stack and give server ability to parse argument
address_type