This is happening with two Arch systems, one is the May image, kernel 3.6.11-11, fully updated offline with Pacman -Su on May 26 and has netctl 1.0-1, the other is with the January image updated to kernel 3.6.11-9 and has netctl 0.8-1. So the problem is from before the big move of /usr/bin /sbin and all that. (I do not want to fully update until the dust settles with upgrades as my main machine is on dialup and the download time is too intense.)
Also note that my Pi could go online in March or April without issue before installing netctl, but not after. So it looks like netctl or my netctl profile is related.
All uncommented lines in file: /etc/netctl/ethernet-static
Code: Select all
Description='A basic static ethernet connection'
Interface=eth0
Connection=ethernet
IP=static
Address=('192.168.xxx.x/24')When I say ethernet goes down, I mean that my ssh session freezes and when I then close my ssh console and try to ssh back in over ethernet, there is "No route to host". I issue the normal ifconfig that usually gets me in, but there is no ethernet to connect with on the Pi end, even though I do see the Pi's LINK light is on.
When I say USB goes down, I mean that I cannot get my usb keyboard to be recognized (so I cannot enter commands or do ctl-alt-del to reboot). And the usb network connection to my zaurus sl5500 goes down as well. When I try to get back in via ssh over usb, get
ssh: connect to address 192.168.xxx.xx port 22: Network is unreachable
The first time this happened on the 3.6.11-9 kernel, ACT light on Pi was on solid for 15 minutes while my ssh session was frozen, and I pulled the power plug. When I checked the logs, the last entries were at the point the ssh session froze, there were no entries during the ensuing 15 minute period during which the time ssh session was frozen, none until after I pulled the plug and restarted the system. But this was with loglevel=2 on one card, and no loglevel declared in cmdline.txt on the other one.
I have a new powered hub with 2.6 mA adapter so can take the Pi online and simultaneously use my USB speakers, so this could be because of issues with the hub except I have an old Arch SD card with the Sept 2012 soft float image on it, and I was able to go online successfully with it for over a half hour with the identical hardware, after these other events.
So, I will set loglevel=6 on both the hard float cards as the logs at loglevel=2 are useless for debugging. And I will test TP1-TP2 before and after the events, just in case power is an issue. But is there anything else I should do or run that can be done before the session freezes, to debug this?
I do not think this is SD card corruption. I immediately ran fsck on both SD cards after the events, and while the journals needed to be restored because of yanking power without a safe shutdown, no other errors were found.
Pi does connect to internet according to logs that look like:
Code: Select all
Jun 8 22:03:34 alarmpi pppd[6533]: Cannot determine ethernet address for proxy ARP
Jun 8 22:03:34 alarmpi pppd[6533]: local IP address 66.81.xx.xx
Jun 8 22:03:34 alarmpi pppd[6533]: remote IP address 69.19.xx.xx
Jun 8 22:03:34 alarmpi pppd[6533]: primary DNS address 64.136.xx.xx
Jun 8 22:03:34 alarmpi pppd[6533]: secondary DNS address 64.136.xx.xx
Jun 8 22:03:34 alarmpi pppd[6533]: Script /etc/ppp/ip-up started (pid 6561)
Jun 8 22:03:35 alarmpi pppd[6533]: Script /etc/ppp/ip-up finished (pid 6561), status = 0x0
Jun 8 22:04:03 alarmpi pppd[6533]: sent [LCP EchoReq id=0x1 magic=0xe279f241]
Jun 8 22:04:04 alarmpi pppd[6533]: rcvd [LCP EchoRep id=0x1 magic=0x0] 
.