On Wed, Aug 31, 2016 at 09:16:00AM -0500, Richard Owlett wrote:
I have Jessie installed on a desktop and a laptop. They both take about 105-110 seconds to boot. Squeeze on the desktop takes ~15 seconds to boot.Most of the delay seems to be in something that states approximately "A start job is running for dev-disk-by\x2duuid?????.device (ttttt}" Where ????? is a long alphanumeric string ttttt is a timer displayThat is followed by two message identified as "Warning" in bright yellow.[they pass by to fast to read].What's going on?
UUID is considered to be the most reliable way to locate a disk, given modern kernels' asynchronous device discovery. It is no longer considered "safe" to assume that /dev/sda is your main hard drive. On small systems, it probably won't change, but if you have several drives, they can start at different times, and the order in which they get enumerated may change. Therefore, Linux now tends to refer to disks by UUID (Universally Unique IDentifier). The boot process now involves various methods for finding that UUID and, as you see above, waiting some considerable time for the device to appear (the device might be dependant on another host, or it might be on a relatively slow bus such as USB or SCSI-1 etc)
Now, there are a number of reasons why you might get the messages above, but one common 'gotcha' is that systemd is more strict about what devices are needed to boot the system. Basically, systemd assumes that, if you've listed a device in /etc/fstab and you've not told it otherwise, then that device is necessary for the system to boot. The two main ways to disabuse it of this notion are to either add the option "noauto", in which case the device will not get mounted at boot but you can still invoke "mount /path/to/mountpoint" once the system is up, or else you can add the option "nofail" which tells systemd that a failure to mount the device should not lead to a failure of the service (and thus booting can continue).
Another reason, and this may well be something you're hitting, is that the UUID doesn't match any devices on the system. UUIDs are, typically, created by mkfs (or mkswap). If you reformat a partition, the UUID will probably change, and your /etc/fstab might be pointing to the old device. If, say, you create filesystems and restore a backup to them, then /etc/fstab can be pointing to UUIDs which no longer exist; if that's the case, then consider using LABEL=something in /etc/fstab instead. UUIDs are reliable, but they are computer generated. Sometimes a maintainer-generated identifier is better.
Are the messages that scroll past preserved anywhere? TIA
-- For more information, please reread.
Attachment:
signature.asc
Description: PGP signature