Hello Camaleón, > > we have a strange problem here at our company: > > > > We have a few linux machines behind a firewall. > > What kind of firewall? Iptables rules or some kind of commercial > appliance?
It's a commercial appliance: Sonicwall NSA3500. > Computers with older kernels are running the same OS version than > machines with newer kernels? I mean, is the kernel version the only > difference between the machines that behave okay or are another > factors that come to play? It's a bit the other way round. These are a few machines that I tested: PT-AGCMLX1 2.6.26-2-686 #1 SMP Wed Nov 4 20:45:37 UTC 2009 PT-CMDEVLX1 2.6.26-2-686 #1 SMP Wed Nov 4 20:45:37 UTC 2009 CMDevLin2 2.6.26-2-686 #1 SMP Wed Nov 4 20:45:37 UTC 2009 CMDevLin3 2.6.32-5-686 #1 SMP Mon Jun 13 04:13:06 UTC 2011 pc-cm-gru-virtlinux1 2.6.32-5-686 #1 SMP Wed May 18 07:08:50 UTC 2011 They all run debian. The older ones (the first group with 2.6.26 kernels) run debian 5.0, the newer ones run debian 6.0. A colleage of mine who didn't arrive yet has an always up to date archlinux with a kernel newer than 2.6.32. Among them are two selfassembled servers, an elder dell quad core server, two dell desktops and two virtual machines (one of them a virtualbox, the other xen). So it's a very heterogeneous group. But: all of them who have a 2.6.32 or newer kernel have the slow-internet-problem. All of them who have a 2.6.26 kernel don't have the problem. The rest of our ~200+ PCs is running windows. They don't have the problem. You can imagine, with how much pressure this issue is handled ;) So for me this boils down to the kernel version in combination with our firewall (or whatever is special here in our company). > > Have you any idea what's wrong? I beg your pardon for my grammar. It's hard to miss that I'm not a native english speaker (blush) > I would try first to remove (not in the sense of "eliminating" but > "bypass") the firewall to discard the problem is generating from > there. Yes, this would be sane. But the company won't switch it off and I don't think we have a way to bypass it :s Thanks for your help, P.S.: Don't miss my next answer to myself ;) -- Markus Grunwald -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20111125085759.54ea0...@pc-cm-gru-virtlinux1.pt.local
