CatBus;194516 Wrote: > > Okay, so it happened again...kinda. The server seems fine, but I'm > trying to transfer files to it and BAM down goes the connection. So I > actually manage to NX into the box (surprise!) and am able to confirm > that / has 15GB free, before BAM, it drops again. Then I just try to > ping the server. I'm getting about 70-80% packet loss. Not 100%. > Previously I was seeing "no route to host" which I think is a totally > different thing. Well, not today.
No route to host on a local network is what some things seem to display when they can't ARP the machine. (ARP is basically an exchange of packets: "Hey, I want to talk to the machine at 1.2.3.4, what is the MAC address for that?" "Oh, that's me: my MAC is in my header!" A machine should not drop ARP requests obviously... > > I notice that my desktop box is the only machine having trouble talking > to the server. My laptop is fine. Desktop problem or just lucky? Hard > to say--yesterday the desktop was talking to the server fine too. > Interesting: my router deaths would usually destroy my internet connection, but local stuff was fine. That the desktop can't talk to the server but the laptop can implies the desktop is the one with an issue. Is it wired or wireless? Does the desktop connect elsewhere? Does rebooting either the desktop or router fix it? > > I check my desktop's /var/log/messages, and there are tons of messages > about packets from "martian sources". Greeeeeeat. So we're back to > the network again, or at least there's a network problem in addition to > the other problem. Many of the things I did next may have done nothing > to solve the problem, but it felt good to do something. > That would imply a network problem. Martian packets are packets that make no sense to the kernel. Like packets from 255.255.255.255, which should never exist. It could be something on the network spewing junk, corrupted junk, or the server is broken and corrupting it as it arrives... but the laptop working implies that isn't the case. > > So either I played with the hardware for long enough for the gremlins > to get a good laugh out of me and scurry away, or something got fixed. > Perhaps totally unrelated to the original problem, but something. Hrrm... or related: memory error? Depending on where bad bits get stuck, you could get totally random stuff: sometimes it would happen to be in a packet buffer and corrupt packets, sometimes it would be in a process and make it crash and die, etc, etc... When you can go without music for a while, grab a copy of memtest86, put it on a floppy (um, or a CD.. depending on vintage of machine), boot from it, and let it run and test memory for a few hours. (The BIOS memory "test" isn't much of a test at all.) Today's symptoms really sound hardwarish more than softwarish. -- snarlydwarf ------------------------------------------------------------------------ snarlydwarf's Profile: http://forums.slimdevices.com/member.php?userid=1179 View this thread: http://forums.slimdevices.com/showthread.php?t=34229 _______________________________________________ unix mailing list [EMAIL PROTECTED] http://lists.slimdevices.com/lists/listinfo/unix
