From: ibuprofin@painkiller.example.tld.invalid   
      
   On Sat, 19 Jan 2013, in the Usenet newsgroup alt.os.linux.mandriva, in article   
   , Adam wrote:   
      
   >Moe Trin wrote:   
      
   [Thinkpad T60]   
      
   >> there is no need for hurry. You don't need it in class yet, right?   
      
   >Nope, looks like I won't need it at all there. (Although spring-term   
   >classes are a great place to pick up anything contagious, ugh.)   
      
   Is that medical, or computer contagions?   
      
   >Meanwhile I made other discoveries about the laptop. When I boot   
   >Mandriva on battery power, it's VERY slow (presumably to conserve the   
   >battery charge), even if I then connect it to AC. I'm sure that's   
   >configurable.   
      
   I believe that's an acpid function. That is _probably_ distribution   
   specific, and there should be a helper tool or menu item for it.   
      
   >Also, its built-in dialup modem is a customized Winmodem.   
   > I thought that might be handy for sending and getting faxes away   
   >from home, but then I've never had a need for that.   
      
   That would also assume you have the appropriate phone cable AND a   
   place to plug in and make the call. I can probably count the number   
   of times I've had access to a public "open" phone jack. On the other   
   hand, neither of our laptops has a modem of any kind, but the little   
   US Robotics 5637 USB modems I have are small enough to carry (about a   
   pack of cigarettes with cables coiled) if needed.   
      
   >> I _thought_ they were getting IPs from the router - and the   
   >> _router_ was configured to hand out specific IPs to specific MACs.   
   >> Yes, that is static addressing, but it's still using DHCP.   
      
   >No, everything has a static address specified in   
   >/etc/sysconfig/network-scripts/ifcfg-whatever . That's because (a) it   
   >worked and (b) I don't recall any recommendations to do it any other   
   >way.   
      
   Oh, OK - I misunderstood your setup. DHCP is the more common setup   
   because it usually works and requires little to no thinking on the   
   part of the person setting up the computer. The technical burden is   
   transferred to the person who has to set up the DHCP server, and that   
   may or may not be a problem.   
      
   >> Because the wireless side of helot does indeed go out and about,   
   >> I'd suggest setting it for DHCP. At home if you're using the   
   >> wireless, I'd suggest setting the router to hand out a "fixed"   
   >> wireless address based on the hardware address associated with   
   >> helot's wireless interface   
      
   >Same thing as the MAC address?   
      
   Yup - but almost certainly a different value from that of eth0. My   
   HP has a 64:31:50: (HP) MAC on eth0 and a 4c:0f:6e: (Happy Peoples   
   Factory No 174) on wlan0.   
      
   >> If you are also using it on Ethernet, either set that with a   
   >> static address, _OR_ set the router to give out a "fixed" address   
   >> based on the MAC address of the Ethernet interface (what is shown by   
   >> "/sbin/ifconfig eth0"). My understanding of your setup was you are   
   >> using one range on the Ethernet (ex: 192.168.1.0/28), and a separate   
   >> range (ex: 192.168.1.16/28) for the wireless.   
      
   >No, because nobody ever suggested that. Only suggestion I remember   
   >was moving the router from 192.168.1.1 (and I learned the router has   
   >to have the lowest used address). Would using separate ranges for   
   >wired and wireless make things easier or better?   
      
   Boy, I'm batting a thousand on your setup. It may make it easier for   
   for router, but if it's acting as a bridge this may not be the case.   
      
   >192.168.1.23 eris.aplomb.invalid eris   
   >192.168.1.24 eris-wired.aplomb.invalid eris-wired   
      
   >and the address ending in 3 is the "usual" one for that device.   
      
   OK, let's start by talking about the routing problem, and we'll assume   
   everything is using Ethernet at the moment. Host A wants to talk to   
   host B and the address is in the 192.168.1.x range appropriate. The   
   routing table would have a single entry for this, looking like   
      
    Kernel IP routing table   
    Destination Gateway Genmask Flags Metric Ref Use Iface   
    192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 655 eth0   
      
   This says everyone in 192.168.1.x is reachable directly on the eth0   
   interface. A wireless only setup would look the same, differing only   
   in the name of the interface. Now let's get complicated and have BOTH   
   interfaces up, and some systems reachable on the Ethernet, some only   
   on wireless. The routing table looks like this now   
      
    Kernel IP routing table   
    Destination Gateway Genmask Flags Metric Ref Use Iface   
    192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 655 eth0   
    192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 0 wlan0   
      
   and maybe you see the problem. How do you know which route to use?   
   They both lead to 192.168.1.x, and have equal Metrics. The Linux   
   kernel knows in this case to use the last route declared/set-up, under   
   the assumption that who-ever did so screwed up with the first route,   
   and corrected that - BUT FAILED TO REMOVE THE ORIGINAL MISTAKEN ONE.   
   So the kernel does you a favor and ignores any/all duplicate routes   
   except the last one - note the "Use" counts and "Tx packets" counts in   
   the "/sbin/ifconfig -a" output.   
      
   There are several ways around this little mess, each more complicated   
   than the other ;-) The solution involves having one system with   
   both interfaces working (the router?) acting as a relay between the   
   two links. I thought you had solved this when you had eris on one   
   link, and stolid or retread on the other - using the DSL router as a   
   transparent bridge. That involves telling the router to act as a   
   bridge and have it do proxy-arp if needed. Stolid would try to talk   
   to the hpclj5 by noting it has a "local" IP address, sending an ARP   
   packet ("who-has 192.168.1.33") and the hpclj5 on hearing the request   
   responding with a "192.168.1.33 is-at 00:40:17:34:ef:40"). If it   
   were to try to talk to eris, AND eris is on the wireless link, eris   
   wouldn't hear the ARP request - and stolid would report "you can't   
   get there from here"... UNLESS the DSL router is doing bridging and/or   
   proxy-arp/forwarding. The router hearing the arp request, AND if   
   it knows eris is up and on wireless (it should check), would reply   
   to stolid saying "192.168.1.33 is-at ds:l-:ha:rd:wa:re" and stolid   
   would be able to talk (with the router silently forwarding packets   
   each way). ANOTHER way to do this is to have the wireless on   
   (example) 192.168.2.x while the Ethernet remains on 192.168.1.x and   
   have the relay listed as the gateway:   
      
    Kernel IP routing table   
    Destination Gateway Genmask Flags Metric Ref Use Iface   
    192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 655 eth0   
    192.168.2.0 192.168.1.1 255.255.255.0 UG 0 0 44 eth0   
      
   and   
      
    Kernel IP routing table   
      
   [continued in next message]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   
|