New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
WFLY-3246 Upgrade JGroups to 3.5.1.Final #6777
Conversation
Linux Build 5119 is now running using a merge of 876fbbd |
Linux Build 5119 outcome was FAILURE using a merge of 876fbbd Build problems:Failed tests detected Failed tests
|
There's seems to be a regression with some aspect of the UDP stack (it's what these 2 failing tests have in common) and ipv6 on Linux. Investigating... |
Linux Build 5121 is now running using a merge of 876fbbd |
Linux Build 5121 outcome was FAILURE using a merge of 876fbbd Build problems:Failed tests detected Failed tests
|
As I suspected - all udp-based tests fail on Linux (ipv6 only). Still investigating... |
Windows Build 275 outcome was FAILURE using a merge of 876fbbd Build problems:Failed tests detected Failed tests
|
Linux Build 5127 is now running using a merge of 876fbbd |
Linux Build 5128 is now running using a merge of 876fbbd |
Linux Build 5129 is now running using a merge of 876fbbd |
Linux Build 5130 is now running using a merge of 876fbbd |
Linux Build 5128 outcome was FAILURE using a merge of 876fbbd Build problems:Failed tests detected Failed tests
|
Linux Build 5127 outcome was FAILURE using a merge of 876fbbd Build problems:Failed tests detected Failed tests
|
Linux Build 5129 outcome was FAILURE using a merge of 876fbbd Build problems:Failed tests detected Failed tests
|
Linux Build 5130 outcome was FAILURE using a merge of 876fbbd Build problems:Failed tests detected Failed tests
|
Linux Build 5139 is now running using a merge of c2e1c75 |
OK - I've figured out what has changed in JGroups 3.5 to cause multicast discovery to break in the CI. In the UDP protocol of JGroups 3.4, multicasts are sent and received via a MulticastSocket, while unicasts are sent and received via a DatagramSocket. On Linux, we would bind the MulticastSocket to the mcast_addr to eliminate cross-talking (see https://issues.jboss.org/browse/JBPAPP-1539), while the DatagramSocket is bound to the configured bind_addr (i.e. the loopback interface). Since this socket is bound to the loopback interface, datagram packets sent to the destination multicast (node-local) address are never received on the MulticastSocket - thus discovery doesn't work. @ctomc @n1hility I suggest we modify the default parameters for the Linux agents on TC to use a valid non-loopback interface. Our testsuite sets the TTL for multicasts to 0, so network chatter shouldn't be an issue. |
Linux Build 5139 outcome was FAILURE using a merge of c2e1c75 Build problems:Failed tests detected Failed tests
|
Linux Build 5161 is now running using a merge of 5702319 |
Linux Build 5162 is now running using a merge of 5702319 |
Linux Build 5162 outcome was FAILURE using a merge of 5702319 Build problems:Process exited with code 143 |
Linux Build 5163 is now running using a merge of 5702319 |
Linux Build 5163 outcome was FAILURE using a merge of 5702319 Build problems:Failed tests detected Failed tests
|
Linux Build 5164 is now running using a merge of 5702319 |
Linux Build 5164 outcome was FAILURE using a merge of 5702319 Build problems:Failed tests detected Failed tests
|
Linux Build 5165 is now running using a merge of 5702319 |
Linux Build 5166 is now running using a merge of 5702319 |
Linux Build 5166 outcome was SUCCESS using a merge of 5702319 |
Linux Build 5169 is now running using a merge of b346323 |
Linux Build 5170 is now running using a merge of b346323 |
Linux Build 5171 is now running using a merge of 8e87a5c |
Linux Build 5170 outcome was FAILURE using a merge of 8e87a5c Build problems:Failed tests detected Failed tests
|
Linux Build 5171 outcome was FAILURE using a merge of 8e87a5c Build problems:Failed tests detected Failed tests
|
Linux Build 5175 is now running using a merge of 8e87a5c |
Linux Build 5175 outcome was FAILURE using a merge of 8e87a5c Build problems:Failed tests detected Failed tests
|
Superseded by #6805 which contains environment changes needed for tests to pass w/ipv6. Closing. |
https://issues.jboss.org/browse/WFLY-3246
Includes protocol stack tweaks, most notably:
Also, WFLY-3445 Allow custom protocol types in jgroups stack
https://issues.jboss.org/browse/WFLY-3445