This is a short story of an interesting fault. But first a small simplified part of my system setup
The problem was this appearing at random times... only lasted for about a few seconds at most for each disruption which could be up to several hours apart.
Tried everything to fix "alice" over about two weeks - nothing helped. Then woke up one morning to find "may" had totally lost internet connectivity. A check showed the e1000 PCI card in "may" had "died". Another is on order and "may" is running using two USB ethernet dongles temporarily while waiting. The majority of internet traffic has been moved over to "alice".
As for my problem with "alice"... Gone
- so "may" was the culprit and no amount of work on "alice" would have fixed it...
BTW "alice" was my maternal grandmother and "may" her twin sister...
http://www.sraellis.tk
----------------------
| |
| may | WAN cable
| main firewall |-- 192.168.0.34 --
| | e1000 PCI dual
LAN | | port port 0
-- 192.168.2.34 ---| Intel Atom |
r8168 onboard | mini-ITX |
| | WAN ADSL2+
| ClearOS 6.9 |-- 192.168.4.34 --
| | e1000 PCI dual
| | port port 1
----------------------
----------------------
| |
| alice | WAN cable
| backup firewall |-- 192.168.0.14 --
| | e1000 PCI dual
LAN | | port port 0
-- 192.168.2.14 ---| Intel Atom |
r8168 onboard | mini-ITX |
| | WAN ADSL2+
| ClearOS 6.9 |-- 192.168.4.14 --
| | e1000 PCI dual
| | port port 1
----------------------
The problem was this appearing at random times... only lasted for about a few seconds at most for each disruption which could be up to several hours apart.
...
May 21 16:08:30 alice kernel: r8168: eth1: link down
May 21 16:08:33 alice kernel: r8168: eth1: link up
May 21 16:09:42 alice kernel: r8168: eth1: link down
May 21 16:09:45 alice kernel: r8168: eth1: link up
...
May 22 12:37:21 alice kernel: ADDRCONF(NETDEV_UP): eth0: link is not ready
May 22 12:37:23 alice kernel: e1000: eth0: e1000_watchdog_task: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX/TX
May 22 12:37:23 alice kernel: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
May 22 12:37:25 alice kernel: eth1: 0xf867c000, 20:cf:30:af:a3:d7, IRQ 28
May 22 12:37:26 alice kernel: ADDRCONF(NETDEV_UP): eth1: link is not ready
May 22 12:37:29 alice kernel: r8168: eth1: link up
May 22 12:37:29 alice kernel: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
May 22 12:37:34 alice kernel: ADDRCONF(NETDEV_UP): eth2: link is not ready
May 22 12:37:37 alice kernel: e1000: eth2: e1000_watchdog_task: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX/TX
May 22 12:37:37 alice kernel: ADDRCONF(NETDEV_CHANGE): eth2: link becomes
ready
...
Tried everything to fix "alice" over about two weeks - nothing helped. Then woke up one morning to find "may" had totally lost internet connectivity. A check showed the e1000 PCI card in "may" had "died". Another is on order and "may" is running using two USB ethernet dongles temporarily while waiting. The majority of internet traffic has been moved over to "alice".
As for my problem with "alice"... Gone
- so "may" was the culprit and no amount of work on "alice" would have fixed it...BTW "alice" was my maternal grandmother and "may" her twin sister...
http://www.sraellis.tk
In Hardware
Share this post:
Accepted Answer
OK - just to close this out as the replacement has arrived from China (ordered 2 so now have a spare).
Original Dual NIC was
The replacement is the same functionally - but different brand, physically slightly smaller and a newer version chipset...
Machine came straight up with the new NICs - so conclude for now the old dual NIC adaptor is dead (will check for sure later in a test machine...)
It's running my web-server amongst other duties - http://www.sraellis.tk/ - let's hope it runs reliably for several more years
Original Dual NIC was
Ethernet controller: Intel Corporation 82546EB Gigabit Ethernet Controller (Copper) (rev 01)
Subsystem: Intel Corporation PRO/1000 MT Dual Port Server Adapter
The replacement is the same functionally - but different brand, physically slightly smaller and a newer version chipset...
Ethernet controller: Intel Corporation 82546EB Gigabit Ethernet Controller (Copper) (rev 03)
Subsystem: Compaq Computer Corporation NC7170 Gigabit Server Adapter
Machine came straight up with the new NICs - so conclude for now the old dual NIC adaptor is dead (will check for sure later in a test machine...)
It's running my web-server amongst other duties - http://www.sraellis.tk/ - let's hope it runs reliably for several more years
Responses (2)
-
Accepted Answer
Possibly PS - but think unlikely...
http://www.sraellis.tk/frame.php?number=34&monitor=voltages shows the average for the two voltages monitored has remained stable - but that of course would not pick up any instantaneous peaks or dips or problems on other voltage rails...
Haven't looked in the machine. It is still running with the 'broken' NICs inside it and wasn't powered off, just added the usb ethernet dongles. It went straight back into operation and hasn't missed a beat since.
We will see what happens when the replacement arrives (coming from overseas - replacements for dual NIC PCI 32-bit gigabyte adaptors are not available here... being Mini-ITX have only one PCI slot).
Please login to post a reply
You will need to be logged in to be able to post a reply. Login using the form on the right or register an account if you are new here.
Register Here »