From mboxrd@z Thu Jan 1 00:00:00 1970 From: ebiederm@xmission.com (Eric W. Biederman) Subject: Re: iproute2: potential upgrade regression with 58a3e827 Date: Fri, 08 Nov 2013 14:42:20 -0800 Message-ID: <87wqkiy0qb.fsf@xmission.com> References: <527D2768.1030403@canonical.com> <87k3gi4lub.fsf@xmission.com> <527D65FB.8000407@canonical.com> Mime-Version: 1.0 Content-Type: text/plain Cc: shemminger@osdl.org, netdev@vger.kernel.org To: Chris J Arges Return-path: Received: from out02.mta.xmission.com ([166.70.13.232]:42356 "EHLO out02.mta.xmission.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758023Ab3KHWmc (ORCPT ); Fri, 8 Nov 2013 17:42:32 -0500 In-Reply-To: <527D65FB.8000407@canonical.com> (Chris J. Arges's message of "Fri, 08 Nov 2013 16:30:19 -0600") Sender: netdev-owner@vger.kernel.org List-ID: Chris J Arges writes: > Fair enough. I really wanted to get a feel if this seemed like an actual > bug, or rather it's behaving like we'd expect. I don't mind writing > code, but I want to know if I'm hunting a bug, or adding a feature. Making it ip netns add work reliably inside of ip netns exec and handling all of the permutations is adding a feature. Understanding why ip netns exec is failing in your specific case sounds like there may be a missing error message. Clearly you get into a context where /var/run/netns/netns_old is no the bind mount we would expect it to be. I saw nothing that should unmake /var/run/netns/nnetns_old as a mount point. So something strange is going on. So you might want to go through and inspect to see what is happening. Perhaps there is a missing error message somewhere. Without understanding why that mountpoint fails to exist I can't say if it is a real bug or if it just something weird caused by using an ip netns exec it was not designed to be used and is known not to be 100% robust in. Eric