[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
Hello. I've got some network issues that I've spent a long time trying to find the reason behind - or even something that causes it. Perhaps someone here has knows some suggestions. I have a fairly large (20+ machines) local network. There are many different machines on the LAN (some Linux boxes, some Windows server ones, some VMs, some hardware appliances) but the gateway machine runs SuSE Linux Enterprise Server 11. The machine is connected to the Internet with a BT 2700 router/hub (and via a BT ASDL business broadband connection). It is this connection that is seeing problems. Basically, the connection occasionally dies and the router cannot be pinged. This lasts between a few seconds to a few minutes. I have some control over the activity on the LAN (which includes the amount of traffic between the LAN and the internet) and reducing it reduces the likeliness of the problems to occur (but currently the problems occur so frequently that I have to reduce the activity to well below what is workable for the problems to disappear almost altogether). The outbound activity on the network consists mainly of http (tcp port 80), https (tcp port 443), smtp (tcp port 25) and a bit of dns (udp port 53) traffic. Not only do I not know what is causing these problems, but other than network traffic I have yet to find something that correlates with the issue. (At the moment, I would be quite happy if I could somehow control this in a workable manner.) Among other things, I have tried the following: - stop most of the outbound SMTP traffic; - stop outbound traffic from any machine on the LAN but the gateway machine; - route DNS traffic elsewhere (two other SLES machines on the LAN have their own outbound connection); - use different router (of the same make); - use different cables; - use a different machine. None of these worked, though the first three had some problem-reducing effect, because they reduce the network traffic too.get to I have also done a lot of extensive monitoring using tcpdump, ipfm etc. to see if something odd happens just before the network goes down, but can't find anything odd. I am pretty certain the network isn't clogged as traffic is a lot higher at other moments. (Also, I would expect to be able to reach the router in such cases.) Perhaps the weirdest thing is that I have done some traffic shaping which reduced the maximum amount of traffic to over the connection. I can see that this does reduce the amount of outbound traffic. However, it does not stop the network from going down occasionally. Erm.. well, that's it. Anyone? Thanks. Martijn. -- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/listfaq