tag:blogger.com,1999:blog-167509166635321801.post6185778234418903927..comments2013-10-17T15:41:18.476-07:00Comments on AAL: Intermittently seeing EADDRNOTAVAIL when calling connectAnonymoushttp://www.blogger.com/profile/08684102779166654965noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-167509166635321801.post-23825360423406491702013-10-17T15:41:18.476-07:002013-10-17T15:41:18.476-07:00FYI: RedHat fixed it in 5.10FYI: RedHat fixed it in 5.10Anonymoushttps://www.blogger.com/profile/08356392349418211996noreply@blogger.comtag:blogger.com,1999:blog-167509166635321801.post-5882038472250699362013-05-19T12:19:08.832-07:002013-05-19T12:19:08.832-07:00Is it possible you are using SO_REUSEADDR? I am u...Is it possible you are using SO_REUSEADDR? I am unable to reproduce the behavior you describe on RH 5.9 (2.6.18-348) x86_64, unless I use SO_REUSEADDR. There was a significant change from 5.8 whereby the kernel's port selector will select any port that is bindable. (See 5.9's patch against inet_csk_get_port.) With SO_REUSEADDR set, this can include ports that are already part of an established connection. If one of these in-use ports is selected for your socket, and you attempt to connect to the exact same server, then EADDRNOTAVAIL will be emitted.Anonymoushttps://www.blogger.com/profile/02842753709117700402noreply@blogger.comtag:blogger.com,1999:blog-167509166635321801.post-38189353410099417442013-05-09T13:26:02.563-07:002013-05-09T13:26:02.563-07:00Andrew, you have indeed paid back some of the help...Andrew, you have indeed paid back some of the help you've received on the web. <br /><br />I encountered this issue, and your entry here was tremendously helpful.Diebold 1063ixhttps://www.blogger.com/profile/01418942527560299861noreply@blogger.comtag:blogger.com,1999:blog-167509166635321801.post-31041203891031735542013-03-05T15:27:43.590-08:002013-03-05T15:27:43.590-08:00Hey Andrew,
Thanks for the post. Seeing very simi...Hey Andrew,<br /><br />Thanks for the post. Seeing very similar behavior on 2.6.32-279.2.1.el6.x86_64. <br /><br />Still investigating to make sure it's not us, but it does seem like kernel-specific behavior.<br /><br />- AlexAnonymoushttps://www.blogger.com/profile/17290111775750077413noreply@blogger.com