zouyu,
Hi jfriesse, Could you please help to review the change ' d2f8e181b6cefb36250e3d9a1baadfbb016e108a' which fix to the problem about 'corosync crashes if network is down' in the github when you have time? According to the discussion before, i use the local socket pair instead of binding '127.0.0.1' if all network interfaces are down.
yep. Sorry. I've actually wanted to put comment to github. I didn't had time to look to your patch (some higher priority things). But I'm expecting to review it next week.
Thanks for your patience Regards, Honza
And below are the tests about the fix: 1. setup 2-nodes cluster, n1 and n2. 2. on n1, run 'testcpg cpg_bm'. and on n2, run 'testcpg cpg_bm' and 'cpgbench' and n1 can receive all the messages from n2. 3. stop all network interfaces on n2 and after a while, n1 has no messages to receive. 4. on n2, at the console of testcpg, test one message and n2 can receive its only message without block. 5. start network on n2. and continue cpgbench again, n1 can continue to receive the messages from n2. Thanks & Best wishes Yu, Zou
_______________________________________________ discuss mailing list discuss@xxxxxxxxxxxx http://lists.corosync.org/mailman/listinfo/discuss
