A machine of mine which is a simple file-server (Dell PowerEdge 430 IIRC) seems to be having some trouble when running multiple NICs.
Initially, I tried to bond two NICs together under the IP address 192.168.1.3. Although the bond worked and both ETH0 and ETH1 were ‘enslaved’ by BOND0, all network operations, such as playing AVI files had dire performance. Dire as in the initial cache took more than a minute to fill (previously took a second or two) and files would stutter and skip while playing.
Prior to the bond, there was a single NIC and everything ran perfectly.
Now I have removed the bond and reset the configuration but left in the 2nd NIC (ETH1) - the idea being that I play media from ETH0 and upload new content to the server via ETH1.
I don’t know what the problem is but it seems that only one NIC is actually doing anything (both have been configured in yast)!
To test I loaded two folders (folder1 and folder2) in Windows both approx 3.4GB in size; I copied these folders from the source to \192.168.1.3 heshare and \192.168.1.4 heshare
I then watched the NICs using ifconfig on the server.
ETH0’s TX/RX was shooting up in MB as expected but ETH1’s TX/RX was creeping up in KB.
Windows said that it would take
13 mins ==> ETH0 - 192.168.1.3
143 mins ==> ETH1 - 192.168.1.4
I would have expected that both ETH0 and ETH1’s TX/RX would be increasing pretty much roughly the same.
For now, I have re-configured the server to only 1 active NIC (ETH0) and am copying the two folders to it and Windows says about 30 minutes to go, which is about “right”).
Does anyone have any ideas or diagnostic tools they can suggest to truly monitor the TX/RX rates of the two cards?
In addition to software monitoring of the transfer rates, such as System
Guard or iftop on the file server, some of the NIC’s I’ve used have LED’s
to indicate when one transfer rate or another have been reached. A server
in my apartment has the power LED of the case connected to some of one of
the NIC’s pins for such a purpose. It’s easier to see than the LED’s on the
back of the case.
badger fruit wrote:
>
> Suse 11.0
>
> A machine of mine which is a simple file-server (Dell PowerEdge 430
> IIRC) seems to be having some trouble when running multiple NICs.
>
> Initially, I tried to bond two NICs together under the IP address
> 192.168.1.3. Although the bond worked and both ETH0 and ETH1 were
> ‘enslaved’ by BOND0, all network operations, such as playing AVI files
> had dire performance. Dire as in the initial cache took more than a
> minute to fill (previously took a second or two) and files would stutter
> and skip while playing.
>
> Prior to the bond, there was a single NIC and everything ran
> perfectly.
>
> Now I have removed the bond and reset the configuration but left in the
> 2nd NIC (ETH1) - the idea being that I play media from ETH0 and upload
> new content to the server via ETH1.
>
> I don’t know what the problem is but it seems that only one NIC is
> actually doing anything (both have been configured in yast)!
>
> To test I loaded two folders (folder1 and folder2) in Windows both
> approx 3.4GB in size; I copied these folders from the source to
> \192.168.1.3 heshare and \192.168.1.4 heshare
>
> I then watched the NICs using ifconfig on the server.
> ETH0’s TX/RX was shooting up in MB as expected but ETH1’s TX/RX was
> creeping up in KB.
>
> Windows said that it would take
> 13 mins ==> ETH0 - 192.168.1.3
> 143 mins ==> ETH1 - 192.168.1.4
>
> I would have expected that both ETH0 and ETH1’s TX/RX would be
> increasing pretty much roughly the same.
>
> For now, I have re-configured the server to only 1 active NIC (ETH0)
> and am copying the two folders to it and Windows says about 30 minutes
> to go, which is about “right”).
>
> DOES ANYONE HAVE ANY IDEAS OR DIAGNOSTIC TOOLS THEY CAN SUGGEST TO
> TRULY MONITOR THE TX/RX RATES OF THE TWO CARDS?
>
> Thank you for reading and hope someone can help!
>
>
Sorry for taking a while to reply; I don’t think the switch as this feature and I wasn’t aware that was a requirement either
I have however found the actual problem!
After running two konsole sessions (one showing iftop, and the other running “tail -f /var/log/messages”) and then re-attempting to copy data to the share I noticed lots of “Device error - restarting SATA link” messages.
I tried an alternative SATA port (two 1TB disks in an LVM) and the problem continued (same message).
The server was built to replace a small-form desktop PC (which was done because the disks were sitting OUTSIDE the machine on top of it lol, no space inside for these disks or additional ones) so the KNOWN WORKING disks were moved over to this new machine.
I moved the disks back to the original machine and now they’re all working OK there are no errors so I can only imagine either the “new” machine didn’t like 2 x TB disks or had a bad motherboard/sata interface.
