ubuntu server network fail
2 views
Skip to first unread message
James Grossmann
Jul 19, 2010, 4:09:15 PM7/19/10
to bismanlug
I'm looking to reconfigure the network on my server system, I moved it
to a new home, in a rack mount case and added a gigabit nic to it (the
motherboard only has 100mbit). However, since the move, I've been
having network problems, it takes me 30 sec to 1 min to get the
password request when I ssh into it. It has been failing at a number
of updates off of the internet--gentoo portage sync and
mythfilldatabase are the ones I've noticed so far-- and when I ping
the external net, I get nothing, not even a response as though it
attempted to ping (ie. how many tries and fails, etc.), either with ip
address or a domain name. I note that dns resolution doesn't seem to
be too good internally either, when I ping a local host (my desktop),
I get the same result, however, when I ping an ip address internally,
it works fine. I'm using my wrt54gs to do dhcp, and I did do
something perhaps out of the ordinary, I assigned the new nic to the
old address of the server, reassigning the old nic to a different
address... After installing the system into the new case, I added the
eth1 (gigabit) to /etc/network/interfaces, which now looks like this:
to a new home, in a rack mount case and added a gigabit nic to it (the
motherboard only has 100mbit). However, since the move, I've been
having network problems, it takes me 30 sec to 1 min to get the
password request when I ssh into it. It has been failing at a number
of updates off of the internet--gentoo portage sync and
mythfilldatabase are the ones I've noticed so far-- and when I ping
the external net, I get nothing, not even a response as though it
attempted to ping (ie. how many tries and fails, etc.), either with ip
address or a domain name. I note that dns resolution doesn't seem to
be too good internally either, when I ping a local host (my desktop),
I get the same result, however, when I ping an ip address internally,
it works fine. I'm using my wrt54gs to do dhcp, and I did do
something perhaps out of the ordinary, I assigned the new nic to the
old address of the server, reassigning the old nic to a different
address... After installing the system into the new case, I added the
eth1 (gigabit) to /etc/network/interfaces, which now looks like this:
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).
# The loopback network interface
auto lo
iface lo inet loopback
# The primary network interface
auto eth0
iface eth0 inet dhcp
auto eth1
iface eth1 inet dhcp
If it were Gentoo, I'd be rocking already, and in some ways I am, I
can do transfers across the gigabit segment of my net at great speed,
even with this problem, but I would really prefer things to work
correctly. I'm just not as comfortable with the ubuntu-server network
setup as I am with Gentoo.
Thanks in advance,
James
Matthew Lengenfelder
Jul 19, 2010, 4:19:40 PM7/19/10
to bism...@googlegroups.com
I'd bet on it being a bad cable that can't quite handle the gigabit speeds. and/or the cable that is being used for the cable modem (if the errors are only happening with traffic outside of your network).
James Grossmann
Jul 19, 2010, 4:35:47 PM7/19/10
to bism...@googlegroups.com
I've got a brand new cat6 cable between the server and the switch, I
suppose I could trade it out... The rest of the computers on the
network go through the gigabit switch and connect to the outside world
just fine.
Just thinking about it, is it a bad thing if the two nics come out on
different switches? I've got a 5-port gigabit switch, to which,
obviously, the new nic is attached, and I've got an old (surplus
cisco) 100mbit switch to which the old nic is attached (which is then
attached to the gigabit switch). I assume that's not a big deal, but
maybe I'm wrong, I mean people set up data centers with hot backups
all the time...
Mentioning that, I've been thinking of doing the "link aggregation" /
bonding thing, I'm really not in need of a faster connection, gigabit
will certainly be adequate for my needs, however, I would like to have
a fallback setup if possible... I've been looking at
https://help.ubuntu.com/community/LinkAggregation and it seems easy.
I am curious which level of bonding I ought to go with as well, the
higher levels seem interesting, but are perhaps overkill.
Thanks,
James
suppose I could trade it out... The rest of the computers on the
network go through the gigabit switch and connect to the outside world
just fine.
Just thinking about it, is it a bad thing if the two nics come out on
different switches? I've got a 5-port gigabit switch, to which,
obviously, the new nic is attached, and I've got an old (surplus
cisco) 100mbit switch to which the old nic is attached (which is then
attached to the gigabit switch). I assume that's not a big deal, but
maybe I'm wrong, I mean people set up data centers with hot backups
all the time...
Mentioning that, I've been thinking of doing the "link aggregation" /
bonding thing, I'm really not in need of a faster connection, gigabit
will certainly be adequate for my needs, however, I would like to have
a fallback setup if possible... I've been looking at
https://help.ubuntu.com/community/LinkAggregation and it seems easy.
I am curious which level of bonding I ought to go with as well, the
higher levels seem interesting, but are perhaps overkill.
Thanks,
James
Eric Wanchic
Jul 19, 2010, 4:36:34 PM7/19/10
to bism...@googlegroups.com
I second, that might be the easiest. Do you have a tester?
Assuming that is not the problem, sorry, I might be a little lost in translation. I have a couple of questions unanswered, but yet assuming everything worked before....
Is the interfaces file you referenced part of an ubuntu firewall/NAT? by the looks of it, it should get a random IP from the Cable (I'll call connection A) and a random IP from the DHCP (from the Wireless DHCP, I call B), yet...that means that the gateway could be random, unless you force connection B an IP based on MAC address. For me, I have DHCP on the same firewall/NAT box.
Plus, if I have the network setup correct, do you have IP masquerading enabled? If not, that will prevent you from getting out into the Internet.
Finally, I assume you do have a DHCP with your cable supplier, or is it a STATIC IP. If it is Static, you might have reset some things and you'll have to call their tech support and ask them to "Press the Button" Not joking. This is what fixes it, and then you'll need to change interfaces to that it is a static IP, of course.
I hope this helps, but again I'm a little lost too :)
Eric
Assuming that is not the problem, sorry, I might be a little lost in translation. I have a couple of questions unanswered, but yet assuming everything worked before....
Is the interfaces file you referenced part of an ubuntu firewall/NAT? by the looks of it, it should get a random IP from the Cable (I'll call connection A) and a random IP from the DHCP (from the Wireless DHCP, I call B), yet...that means that the gateway could be random, unless you force connection B an IP based on MAC address. For me, I have DHCP on the same firewall/NAT box.
Plus, if I have the network setup correct, do you have IP masquerading enabled? If not, that will prevent you from getting out into the Internet.
Finally, I assume you do have a DHCP with your cable supplier, or is it a STATIC IP. If it is Static, you might have reset some things and you'll have to call their tech support and ask them to "Press the Button" Not joking. This is what fixes it, and then you'll need to change interfaces to that it is a static IP, of course.
I hope this helps, but again I'm a little lost too :)
Eric
James Grossmann
Jul 19, 2010, 5:19:15 PM7/19/10
to bism...@googlegroups.com
The interfaces file is part of my server, the nat/firewall is all
being taken care of by the linksys dd-wrt router, to which I have only
the broadband and a connection to the gigabit switch connected, it is
supplying a static dhcp to each mac address. From the gigabit switch
I have connected my desktop, the gigabit nic in the server and the
cisco 100mbit switch. I have the 2nd (original) nic in the server
attached to the cisco switch.
I am not certain that I have IP masquerading going, I think I do
through my router, but it's been a while since I set it up, I thought
that was for if I wanted to use my computer as a router, which I am
not doing. I have been using the router to assign internal ips
(192.168.1.x) for a couple of years now, with no problems. The IP
from my supplier is dynamic and I have been using DHCP, so I'm not in
problems there.
What I changed mainly was that I installed the cisco switch
(connecting a couple things to it) and added the gigabit nic to the
server (specifying the old ip to that nic), other than that, the
system has been working fine as it is for years now. I just wanted a
bit more speed...
Thanks,
James
being taken care of by the linksys dd-wrt router, to which I have only
the broadband and a connection to the gigabit switch connected, it is
supplying a static dhcp to each mac address. From the gigabit switch
I have connected my desktop, the gigabit nic in the server and the
cisco 100mbit switch. I have the 2nd (original) nic in the server
attached to the cisco switch.
I am not certain that I have IP masquerading going, I think I do
through my router, but it's been a while since I set it up, I thought
that was for if I wanted to use my computer as a router, which I am
not doing. I have been using the router to assign internal ips
(192.168.1.x) for a couple of years now, with no problems. The IP
from my supplier is dynamic and I have been using DHCP, so I'm not in
problems there.
What I changed mainly was that I installed the cisco switch
(connecting a couple things to it) and added the gigabit nic to the
server (specifying the old ip to that nic), other than that, the
system has been working fine as it is for years now. I just wanted a
bit more speed...
Thanks,
James
James Grossmann
Jul 19, 2010, 5:33:42 PM7/19/10
to bism...@googlegroups.com
Interesting thing to note I ran an ifconfig, and note that the gigabit
device is eth1, getting virtually no traffic... I did replace the
cable, no difference...
device is eth1, getting virtually no traffic... I did replace the
cable, no difference...
ifconfig -a
eth0 Link encap:Ethernet HWaddr xx:xx:xx:xx:xx:xx
inet addr:192.168.1.7 Bcast:192.168.1.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:55927713 errors:0 dropped:0 overruns:0 frame:0
TX packets:61424990 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:1447437214 (1.4 GB) TX bytes:3170178790 (3.1 GB)
Interrupt:23 Base address:0x2000
eth1 Link encap:Ethernet HWaddr xx:xx:xx:xx:xx:xx
inet addr:192.168.1.1 Bcast:192.168.1.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:40386 errors:0 dropped:0 overruns:0 frame:0
TX packets:3956 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:13980573 (13.9 MB) TX bytes:485676 (485.6 KB)
lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:8092 errors:0 dropped:0 overruns:0 frame:0
TX packets:8092 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:4139514 (4.1 MB) TX bytes:4139514 (4.1 MB)
Matthew Lengenfelder
Jul 19, 2010, 6:10:52 PM7/19/10
to bism...@googlegroups.com
If your configuration is the same, then it's not that. If you've added a switch, then the switch has changed and likely some cables have changed. It's more likely the cables than the switch(I've never had a switch that's been the cause of intermittent problems like you describe) I have had cables that have been the cause of intermittent problems. I had one cable in particular that worked fine with a 10/100 switch but when used with a gigabit switch it just randomly drops packets. If it were my system I'd try to narrow it down. Is it only the one computer that is having problems - if so then it's probably the cable going to that one computer. If it's multiple computers, but only when accessing the internet through them, or accessing them from the internet, then it's probably the cable going from the cablemodem to the switch. Buy/Make a new cable, verify that it works and start swapping out old cables with that one cable and homefully at some point the problem goes away. If you have a cable tester you can try that too ... but my one cable tested fine, but didn't work at gigabit speeds.
Collin Scull
Jul 19, 2010, 6:29:09 PM7/19/10
to bism...@googlegroups.com
It might be caused by having two interfaces on the same 192.168.1.0/24
subnet. I have had similar issues when I connect using wireless and
wired at the same time in the office or I leave my office vpn on when I
come in. In both cases I have multiple routes to the same subnet and
packets going in on one interface have responses going out on the other
interface.
subnet. I have had similar issues when I connect using wireless and
wired at the same time in the office or I leave my office vpn on when I
come in. In both cases I have multiple routes to the same subnet and
packets going in on one interface have responses going out on the other
interface.
Collin
Dan Falconer
Jul 20, 2010, 9:22:24 AM7/20/10
to bism...@googlegroups.com
I would suggest unplugging/disabling all interfaces but the one you want to troubleshoot (eth1); eliminating extra interfaces eliminates extra variables. Once you're down to just that interface, work on it from there. The SSH problem might be a DNS issue: if there's no local domain, or no entry for the connecting IP, the destination SSH server will wait for DNS to timeout before proceeding.
I would guess the update issues are caused by having two cards. If one is sending the response and the other is receiving, something might be getting routed wrong. One other thing to check is to make sure that the MAC addresses are not the same; if at some point you did something to clone the MAC of another card for any reason, the network is the part that's really confused: it will have some "ARP" cache or something like that that's causing traffic to go... wonky. I read somewhere something about NMAP fixing such network issues, so you may want to google for that (it may have been something posted on the LUG too).--
Best Regards,
Dan Falconer
http://www.CrazedSanity.com || http://www.Buzzkill.org
I would guess the update issues are caused by having two cards. If one is sending the response and the other is receiving, something might be getting routed wrong. One other thing to check is to make sure that the MAC addresses are not the same; if at some point you did something to clone the MAC of another card for any reason, the network is the part that's really confused: it will have some "ARP" cache or something like that that's causing traffic to go... wonky. I read somewhere something about NMAP fixing such network issues, so you may want to google for that (it may have been something posted on the LUG too).
Best Regards,
Dan Falconer
http://www.CrazedSanity.com || http://www.Buzzkill.org
Reply all
Reply to author
Forward
0 new messages