On Fri, 14 Jun 2019, Andreas Schwab wrote: > On Jun 14 2019, Finn Thain <[email protected]> wrote: > > > On Thu, 13 Jun 2019, Andreas Schwab wrote: > > > >> On Jun 13 2019, Finn Thain <[email protected]> wrote: > >> > >> > IMHO these log messages are a bit silly -- > >> > > >> > [ **] (3 of 3) A start job is running for /dev/sda4 (1min 5s / 1min > >> > 36s) > >> > ... > >> > [ TIME ] Timed out waiting for device /dev/sda4. > >> > [DEPEND] Dependency failed for /dev/sda4. > >> > [DEPEND] Dependency failed for Swap. > >> > ... > >> > Activating swap /dev/sda4... > >> > ... > >> > [ 450.650000] Adding 524284k swap on /dev/sda4. Priority:-2 extents:1 > >> > across:524284k FS > >> > ... > >> > [ OK ] Activated swap /dev/sda4. > >> > >> That just means that the timeout is maybe too short, > > > > Adding swap isn't that complicated. > > It's not about adding swap, it's about adding the device node. You > cannot run swapon until the device node exists. >
Being that this is devtmpfs, the kernel creates the device node then notifies udev, right? > > The device isn't late, swapon() is late (compared to Debian 3 or Debian > > 4). > > No, the device node shows up late. That's a user-space operation, > carried out by udev. > I guess udev (which is part of systemd) took too long to notify the systemd unit? Or the kernel took too long to notify udev? -- > Andreas. > >
