Problem with NSA325V2 Ethernet link up/down

I just installed debian on my NSA325v2. Before that I have been using ArchLinux, but I wanted to use openmediavault.
Since a few weeks I noticed connection drops to the NAS. After looking at the dmesg log, there seems to be an error with the ethernet connection:

Quote
[ 977.263477] mv643xx_eth_port mv643xx_eth_port.0 eth0: link down
[ 977.269739] mv643xx_eth_port mv643xx_eth_port.0 eth0: link up, 1000 Mb/s, full duplex, flow control disabled
[ 977.279665] mv643xx_eth_port mv643xx_eth_port.0 eth0: link up, 1000 Mb/s, full duplex, flow control disabled
[ 1037.313328] mv643xx_eth_port mv643xx_eth_port.0 eth0: link down
[ 1037.319590] mv643xx_eth_port mv643xx_eth_port.0 eth0: link up, 1000 Mb/s, full duplex, flow control disabled
[ 1037.329519] mv643xx_eth_port mv643xx_eth_port.0 eth0: link up, 1000 Mb/s, full duplex, flow control disabled
[ 1097.113418] mv643xx_eth_port mv643xx_eth_port.0 eth0: link down
[ 1097.119401] mv643xx_eth_port mv643xx_eth_port.0 eth0: link up, 1000 Mb/s, full duplex, flow control disabled

I had these problems with ArchLinux as well, but I wanted to see if debian could fix them. However, the errors are still there. I already tried to use a different cable and also a different switch, but that didn't help. Can this be hardware related? Or is it something with the kernel?



Edited 1 time(s). Last edit at 01/14/2017 04:16PM by bodhi.

hensur,

My NSA325v2 has been running without this problem continously for a few years since I've bought it . I think it is not kernel related.

Can you post serial console log, or dmesg log?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

It has been for me as well. At least until one month ago.
I attached the dmesg and serial console log.

I think something in your rootfs that caused this. I think because you are running OpenMediavault. Notice the link is down and back up at every 60 seconds.

My suggestion:

dmesg -T

And then look at the system log to see if you can tell what happens periodically at the same time the link is down that shown in dmesg. Could be some periodic process in OpenMediavault that caused this effect? e.g. timing out on something and try to reset the link.

Could be a bug in OpenMediavault that the new kernel have triggered.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Ok, I found the issue. I don't know why I didn't think of it earlier.
I recently set up check_mk to monitor my network and also deployed the agent to the nas. The agent executes several commands on the clients and sends the output to the monitoring server.
I noticed that everytime I execute "ethtool eth0" the link drops. This is exactly the same command check_mk executes every minute on my nas.
However, on the other hosts I have no such problems, although ethtool is executed there as well.
Is it something with the NIC that causes it to drop the link whenever ethtool is executed?
For now, I disabled the ethtool check and the problems are gone. Thanks for your help!

I just tried to mount some nfs shares and I got this in dmesg:

Quote
[Mon Jan 16 12:42:15 2017] ------------[ cut here ]------------
[Mon Jan 16 12:42:15 2017] WARNING: CPU: 0 PID: 7017 at drivers/net/ethernet/marvell/mv643xx_eth.c:1133 txq_reclaim+0x1bc/0x1e4
[Mon Jan 16 12:42:15 2017] Modules linked in: dm_mod binfmt_misc evdev marvell_cesa des_generic orion_wdt md_mod kirkwood_thermal uio_pdrv_genirq uio gpio_keys fuse ipv6 autofs4 sg sd_mod uas usb_storage sata_mv libata fixed scsi_mod
[Mon Jan 16 12:42:15 2017] CPU: 0 PID: 7017 Comm: sshd Tainted: G W 4.9.0-kirkwood-tld-1 #1
[Mon Jan 16 12:42:15 2017] Hardware name: Marvell Kirkwood (Flattened Device Tree)
[Mon Jan 16 12:42:15 2017] [<c0016a98>] (unwind_backtrace) from [<c0012b58>] (show_stack+0x10/0x14)
[Mon Jan 16 12:42:15 2017] [<c0012b58>] (show_stack) from [<c001f580>] (__warn+0xd0/0xfc)
[Mon Jan 16 12:42:15 2017] [<c001f580>] (__warn) from [<c001f650>] (warn_slowpath_null+0x1c/0x24)
[Mon Jan 16 12:42:15 2017] [<c001f650>] (warn_slowpath_null) from [<c049eba0>] (txq_reclaim+0x1bc/0x1e4)
[Mon Jan 16 12:42:15 2017] [<c049eba0>] (txq_reclaim) from [<c049f3b8>] (mv643xx_eth_poll+0x2c4/0x5f0)
[Mon Jan 16 12:42:15 2017] [<c049f3b8>] (mv643xx_eth_poll) from [<c0539ec8>] (net_rx_action+0x11c/0x3a8)
[Mon Jan 16 12:42:15 2017] [<c0539ec8>] (net_rx_action) from [<c0023938>] (__do_softirq+0x134/0x300)
[Mon Jan 16 12:42:15 2017] [<c0023938>] (__do_softirq) from [<c0023dac>] (irq_exit+0x90/0xf8)
[Mon Jan 16 12:42:15 2017] [<c0023dac>] (irq_exit) from [<c005cca4>] (__handle_domain_irq+0x98/0xcc)
[Mon Jan 16 12:42:15 2017] [<c005cca4>] (__handle_domain_irq) from [<c0009438>] (orion_handle_irq+0x70/0x88)
[Mon Jan 16 12:42:15 2017] [<c0009438>] (orion_handle_irq) from [<c00139f4>] (__irq_usr+0x54/0x80)
[Mon Jan 16 12:42:15 2017] Exception stack(0xc6ca1fb0 to 0xc6ca1ff8)
[Mon Jan 16 12:42:16 2017] 1fa0: 8091e7dc 80936dd0 80936dd0 00000000
[Mon Jan 16 12:42:16 2017] 1fc0: 00001028 8091e7dc beeaa230 00001028 beeaa2cc 7f6e4eb4 7f6e36c8 7f6e36c8
[Mon Jan 16 12:42:16 2017] 1fe0: 00000000 beeaa210 7f65a5c8 7f64fc5c 60000010 ffffffff
[Mon Jan 16 12:42:16 2017] ---[ end trace e814e69f24c1f467 ]---
[Mon Jan 16 12:42:16 2017] ------------[ cut here ]------------

It just happend when I started the nfs mount on another server, until then everything was fine.

I could fix this as well now. The problem was a MTU size of 9000 on the nfs client. I tested the nfs mounts on another pc and everything was fine, so I changed the mtu on the other client as well.

hensur,

> I could fix this as well now. The problem was a
> MTU size of 9000 on the nfs client. I tested the
> nfs mounts on another pc and everything was fine,
> so I changed the mtu on the other client as well.

ethtool should not cause the link to drop. This was the real problem then.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)