All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
X9SCL-F-O issue
This isn't probably the best place to post this, but I have confidence that the members of LET can help me.
I'm currently facing an issue with my servers where it is only bursting to 100mbits, instead of the 1GBit/s that it should. This is happening on all of my servers that use the X9SCL-F-O. I can confirm that the issue is not switch or cable related (my custom-built dells are bursting just fine). I am stuck on what to do, here's my output of ethtool (I am running CentOS 6.3 min):
[root@MMCS ~]# ethtool eth1 Settings for eth1: Supported ports: [ TP ] Supported link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Full Supports auto-negotiation: Yes Advertised link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Full Advertised pause frame use: No Advertised auto-negotiation: Yes Speed: 100Mb/s Duplex: Full Port: Twisted Pair PHYAD: 2 Transceiver: internal Auto-negotiation: on MDI-X: on Supports Wake-on: pumbg Wake-on: g Current message level: 0x00000007 (7) Link detected: yes
(eth0 is off because it is connected to a failover switch, which isn't ready for use yet)
Any ideas?
Comments
What OS?
Have you tried
ethtool -S eth1 speed 1000 duplex full autoneg on
edit, GAHHHH @Patrick beat me to it
It makes the nic stop working. Here's ethtool's output:
http://mcserve.rs/Qc2f
0.o
http://mcserve.rs/2WYr
Switch port autoneg disabled maybe?
nm, you discarded that already
Nope, on the old dells, I have it working normally:
http://mcserve.rs/fc1P
Set both the switch-port and the server side to 1000/Full as for some reason you aren't auto-negotiating correctly.
What type of switch?
I am using this: http://www.neweggbusiness.com/Product/Product.aspx?Item=N82E16833127336
The switch says the nic is down when it is forced to use 1000mbits
http://mcserve.rs/3NRq
It sounds like you either have bad cables or a nasty switch.
Typically I would recommend avoiding DLink, NetGear, or Linksys (Now Cisco SmallBusiness) when in use in a server environment and not SOHO. We very frequently see duplex / negotiation issues with these "budget" switches and clients reporting similar problems as yours.
What type of cable are you using, first make sure it's a normal cat5e or cat6. There could be a problem with the cable that isn't an issue @ 100mbps but that could cause it to not link at gig.
Do you have another gig switch to test with and confirm?
I've seen the same issues with the X9SCL on high end switch hardware..everyone on WHT seems to have indicated a driver update will fix it.
Best of luck to you
You should test the last Intel nic driver from elrepo.
What @chriscm & @fileMEDIA said. Here's what I do on my X9SCL+-F boards.
and add
pcie_aspm=off e1000e.IntMode=1,1
in grub.conf in the kernel line.tar -zxf e1000e-*
cd e1000e-*/src
make CFLAGS_EXTRA=-DDISABLE_PCI_MSI CFLAGS_EXTRA=-DE1000E_NO_NAPI install
Doesn't work, but the port is using half duplex instead of full now. Forcing it to use 1000mbits on both the switch and the server does not work either.
I had an issue with something I thought was related to this so I did a bit of research actually and have seen your problem mentioned a lot. While this ended up not being the problem I had, this may be of some help to you:
http://www.doxer.org/learn-linux/resolved-intel-e1000e-driver-bug-on-82574l-ethernet-controller-causing-network-blipping/
Which has the same info in it about using the different wireless driver posted here:
http://lists.centos.org/pipermail/centos/2011-September/118200.html
This also has some of what was said above but the post on doxer.org goes a bit farther in depth and some of it may help you. The part after "change kernel parameter" and "change NIC parameters(you should add these lines to /etc/rc.local)" may help you the most after you change the driver from the default.
Also adding: http://bugs.centos.org/view.php?id=5034 has some info too, may be the same as the other, just adding if it helps (did not read it all)
@mnpeep Try:
1) rpm -Uvh http://elrepo.org/elrepo-release-6-4.el6.elrepo.noarch.rpm
2) yum install kmod-e1000e.x86_64
3) Add pcie_aspm=off to grub.conf
4) Reboot
http://www.doxer.org/learn-linux/resolved-intel-e1000e-driver-bug-on-82574l-ethernet-controller-causing-network-blipping/
http://lists.centos.org/pipermail/centos/2011-September/118200.html
2) yum install kmod-e1000e.x86_64
3) Add pcie_aspm=off to grub.conf
4) Reboot
Both aren't working. Sorry for not replying earlier, I was testing some things at the DC with the switch. The switch isn't the problem, I have set eth0 to feed off another switch (Layer 1 instead of Layer 2), and I am still having the same issue.
Layer 1 must be some new kind of switch i've never heard of before
@rds100 By level 1, I meant the network level. The switch is fully unmanaged, and doesn't have any IP address.
That still Level 2 since it learns / switches based on MAC addresses. Level 1 would be at the electrical level - i.e. an etherhet hub, but i don't think these are made any more and probably can be found only in museums.
Why is everyone suggesting putting the most effort possible into this when ethtool is showing the link at 100mbit? Someone needs to log into the switch and force the port to 1000mbit or try another port
Done, on both the L2 and L1 switches. Here's the output (the port stops working):
http://mcserve.rs/Qc2f
Exactly: http://mcserve.rs/RZNX
Why does your Netgear hace GBIC? How old is that thing?
That is OLD. I don't use it, I just have it laying around in my rack. I'm going to take it home once my second DLink comes in.
Have you tested with another ethernet cable? For gigabit all 8 wires of the UTP cable must me crimped. If one wire doesn't make good contact - the connection will fall back to 100Mbps.
http://mcserve.rs/Qc2f
Tested with a known good cable/port? Install knod-e1000e from elrepo unless this is an ovz kernel, then compile and update to bios 2.0b
Well I'll be... it was due to a bad set of cables. Never seen this before. Now, I'm only going to buy pre-made cat5e cables.
@RyanD, you just got a life point.
SOLVED.
Wow all this effort and it turned out to be a god damn cable!