ODROID-XU3 won't boot after running for a week [SOLVED]

[bmentink]

Hi Guys,

I have been using Arch on my XU3 as a server and it has behaved very well until now.
I went to reboot the server and it hangs in boot when it is doing the "ioboard_keyled_probe" as below:
$this->bbcode_second_pass_code('', '[ 8.751258] [c4] INA231 3-0040: I2C write error: (-6) reg: 0x0
[ 8.751268] [c4] INA231 3-0040: ============= Probe INA231 Fail! : sensor_arm (0xFFFFFFFA) =============
[ 8.751875] [c4] INA231 3-0041: I2C write error: (-6) reg: 0x0
[ 8.751884] [c4] INA231 3-0041: ============= Probe INA231 Fail! : sensor_mem (0xFFFFFFFA) =============
[ 8.752453] [c4] INA231 3-0044: I2C write error: (-6) reg: 0x0
[ 8.752462] [c4] INA231 3-0044: ============= Probe INA231 Fail! : sensor_g3d (0xFFFFFFFA) =============
[ 8.752986] [c4] INA231 3-0045: I2C write error: (-6) reg: 0x0
[ 8.752991] [c4] INA231 3-0045: ============= Probe INA231 Fail! : sensor_kfc (0xFFFFFFFA) =============
[ 8.753085] [c4] i2c-core: driver [INA231] using legacy suspend method
[ 8.753089] [c4] i2c-core: driver [INA231] using legacy resume method
[ 8.838113] [c4] [d:0, c:1] s5p_mfc_open:1725: MFC open completed [0:2] dev = dddff010, ctx = de323800
[ 8.848854] [c4] [d:0, c:0] s5p_mfc_release:1883: MFC driver release is called [0:2], is_drm(0)
[ 8.858975] [c4] [d:0] s5p_mfc_release:2037: mfc driver release finished [0:1], dev = dddff010
[ 8.868975] [c6] [d:0, c:1] s5p_mfc_release:1883: MFC driver release is called [0:1], is_drm(0)
[ 8.879099] [c6] s5p_mfc_dev_release:1801: power off
[ 8.885847] [c6] [d:0] s5p_mfc_release:2037: mfc driver release finished [0:0], dev = dddff010
[ 9.057420] [c7] samsung-pinctrl 14010000.pinctrl: pin gpb3-2 already requested by 12c70000.i2c; cannot claim for ioboard_key_led.12
[ 9.070787] [c7] samsung-pinctrl 14010000.pinctrl: pin-209 (ioboard_key_led.12) status -22
[ 9.075310] [c4]
=================== ioboard_bmp180_probe FAIL! ===================

[ 9.075570] [c4] i2c-core: driver [ioboard-bmp180] using legacy suspend method
[ 9.075580] [c4] i2c-core: driver [ioboard-bmp180] using legacy resume method
[ 9.091117] [c4] ioboard-bh1780 2-0029: I2C write byte error: data=0x8b
[ 9.091122] [c4]
=================== ioboard_bh1780_probe FAIL! ===================

[ 9.091456] [c4] i2c-core: driver [ioboard-bh1780] using legacy suspend method
[ 9.091460] [c4] i2c-core: driver [ioboard-bh1780] using legacy resume method
[ 9.154780] [c7] samsung-pinctrl 14010000.pinctrl: could not request pin 209 on device samsung-pinctrl
[ 9.165282] [c7] ioboard-keyled ioboard_key_led.12: Error applying setting, reverse things back
[ 9.165444] [c7] of_get_named_gpio_flags exited with status 29
[ 9.165455] [c7] of_get_named_gpio_flags exited with status 30
[ 9.165467] [c7] of_get_named_gpio_flags exited with status 22
[ 9.165480] [c7] of_get_named_gpio_flags exited with status 18
[ 9.165497] [c7] of_get_named_gpio_flags exited with status 31
[ 9.165523] [c7] of_get_named_gpio_flags exited with status 28
[ 9.165585] [c7] of_get_named_gpio_flags exited with status 19
[ 9.165670] [c7] of_get_named_gpio_flags exited with status 209
[ 9.165759] [c7] of_get_named_gpio_flags exited with status 24
[ 9.166307] [c7]
=================== ioboard_keyled_probe ===================

[ 97.592151] [c0] smsc95xx 1-1.1:1.0 eth0: hardware isn't capable of remote wakeup
[ 97.604970] [c0] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 99.697368] [c0] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 99.709661] [c0] smsc95xx 1-1.1:1.0 eth0: link up, 10Mbps, half-duplex, lpa 0x0020')

Anyone have any idea's on how to recover this:
It does timeout after a couple of min's into diagnostic mode and I can login as root ... but there is no DNS or sshd so can't access the network or be accessed, but I don't know what to repair anyway.

Maybe a network issue? As that is what it is trying to do next .. maybe an upgrade broke it?

EDIT: I tried doing dhcpcd eth0 and it get's an ip, but then complains about a missing /etc/resolve.conf, if I try reading that file, it says it cannot be read ....... is it corrupt?

Thanks,

[moonman]

Did you disable systemd-networkd before enabling dhcpcd?

[bmentink]

Hi,

Yes I did.

I discovered that the file (/etc/resolve.conf) was corrupted. I deleted it, re-created it and added "nameserver 192.168.0.1" (my router) to it and now dhcpcd eth0 works and I can use the network.

However my boot problem still exists, I suspect it is a systemd problem after reading the forum, as a number of people have had issues with the latest version and have solved by downgrading .. (or I have more corrupted file in /etc) ..

I can't downgrade from my system, as my pacman cache was cleaned a while ago

How can I downgrade from the repo?

[bmentink]

Solved it. Seems a couple of fstab entries for two USB drives stopped the boot if the drives were not present.
Looks like fstab entries and systemd units don't co-exist very well
I installed udevil/devmon to handle my drives, at least there is a systemd unit for that. It works.

Cheers

[moonman]

There are a couple of options you can pass to make it boot. All is in archwiki. It boots in emergency mode without the drives otherwise