| 1 |
>>>> I have this: |
| 2 |
>>>> |
| 3 |
>>>> # dmesg | grep enp |
| 4 |
>>>> [ 4.297862] systemd-udevd[659]: renamed network interface eth0 to enp0s20u2u1 |
| 5 |
>>>> [ 4.778289] systemd-udevd[660]: renamed network interface eth0 to enp0s20u2u2 |
| 6 |
>>>> [ 6.496193] ax88179_178a 3-2.1:1.0 enp0s20u2u1: ax88179 - Link status is: 1 |
| 7 |
>>>> [ 7.905393] ax88179_178a 3-2.2:1.0 enp0s20u2u2: ax88179 - Link status is: 1 |
| 8 |
>>>> # |
| 9 |
>>>> |
| 10 |
>>>> That doesn't tell us when the network initscripts tried and failed to |
| 11 |
>>>> start but this from /var/log/messages/everything/current shows the |
| 12 |
>>>> first time in the boot sequence that a dependent service failed to |
| 13 |
>>>> start because of the networking failure so it should be before this: |
| 14 |
>>>> |
| 15 |
>>>> [kernel] [ 0.787433] serio: i8042 AUX port at 0x60,0x64 irq 12 |
| 16 |
>>>> [/etc/init.d/unbound] ERROR: cannot start unbound as net.enp0s20u2u1 |
| 17 |
>>>> would not start |
| 18 |
>>>> [kernel] [ 0.792081] rtc_cmos 00:04: alarms up to one month, y3k, |
| 19 |
>>>> 242 bytes nvram, hpet irqs |
| 20 |
>>>> |
| 21 |
>>> |
| 22 |
>>> Yeah, so I think the kernel is detecting your network card after udev |
| 23 |
>>> has already started. |
| 24 |
>>> |
| 25 |
>>> One interesting experiment would be to delay the boot process to allow |
| 26 |
>>> the kernel additional time to detect devices. Adding rootdelay=10 to |
| 27 |
>>> your kernel command line should do the trick, unless you are using |
| 28 |
>>> some broken initramfs. |
| 29 |
>> |
| 30 |
>> |
| 31 |
>> I tried that and it works great which I think confirms our suspicions |
| 32 |
>> that the kernel is detecting my network cards after udev has already |
| 33 |
>> started. If I remove rootdelay=10 and I do this: |
| 34 |
>> |
| 35 |
>> # ln -s /dev/null /etc/udev/rules.d/90-network.rules |
| 36 |
>> |
| 37 |
>> the network interfaces fail to come up which is the same thing I've |
| 38 |
>> experienced with rc_hotplug="net.*". |
| 39 |
>> |
| 40 |
> |
| 41 |
> Yeah, so this is not solvable using service dependencies. You will |
| 42 |
> either need to make that boot delay permanent, or rely on the hotplug |
| 43 |
> functionality to start the net.en* services. In the latter case, you |
| 44 |
> should remove them from the default runlevel. |
| 45 |
|
| 46 |
|
| 47 |
Was the 10-second boot delay based on anything in particular or can I |
| 48 |
try a lower delay like 5 seconds? It's tricky to get the machine back |
| 49 |
when I lose it otherwise I would just test it myself. |
| 50 |
|
| 51 |
Would it make sense for me to submit a feature request for network |
| 52 |
interfaces to wait until all USB devices have been initialized before |
| 53 |
starting (or something like that)? |
| 54 |
|
| 55 |
|
| 56 |
> You may want to define rc_need="!net" to prevent init scripts that |
| 57 |
> "need net" from automatically starting the net.* services. For most |
| 58 |
> services this is fine, but it will obviously break things like ntpdate |
| 59 |
> which actually need a usable network connection. |
| 60 |
|
| 61 |
|
| 62 |
I don't follow this. Doesn't hotplug need to be able to start the |
| 63 |
net.* services in order for that solution to work? |
| 64 |
|
| 65 |
- Grant |
Archives