Ubuntu Acknowledges Boot Speed Problem

[Yfrwlf]

Ext4 is a dog.

Canines are awesome, so EXT4 is awesome too?

I'm looking forward to BTRFS to prevent bit rot and keep my data pristine and flawless.

[chenxiaolong]

How will encrypted root partitions be mounted if there's no initrd (and therefore, no cryptsetup)? Or can the initrd be generated manually?

[bug77]

I guess that's one way of handling the power regressions: add a bigger problem and promise you'll fix it, hoping people will forget about the former. I know, power regressions are not Ubuntu's fault, but as one of the bigger distros based on the Linux kernel, there's nothing stopping them from lending a hand and/or providing a bit of feedback every now and then.

@DaemonFC: I have also used XFS trouble-free. But then I messed up the FAT and found out that there's a very thin selection of recovery software for XFS. Now I use XFS on the system partition and EXT on /home.

[LinuxID10T]

As always, who gives a fuck about boot speed? Unless I update the kernel I don't reboot so even if it took 10 mins to boot it wouldn't be a big deal and if you're running a "must always be on" server then why aren't you staggering your systems so that you aren't down regardless?

Its like the flicker free boot, was it necessary for anyone or anything? Does the flicker damage the monitor? If thats the case then why do I have a few CRTs with more then a decade of service still being used with no perceptible wear? If it's damaging to LCDs it must only effect craptacular models since my 5 year old Samsung LCDs have yet to fail me and the screen in my 2002 iBook still looks as good as the day I bought it.

I'm pretty certain it isn't a question of hardware damage, it is a question of polish, and fit and finish.

[Kivada]

As always, who gives a fuck about boot speed? Unless I update the kernel I don't reboot so even if it took 10 mins to boot it wouldn't be a big deal and if you're running a "must always be on" server then why aren't you staggering your systems so that you aren't down regardless?

Its like the flicker free boot, was it necessary for anyone or anything? Does the flicker damage the monitor? If thats the case then why do I have a few CRTs with more then a decade of service still being used with no perceptible wear? If it's damaging to LCDs it must only effect craptacular models since my 5 year old Samsung LCDs have yet to fail me and the screen in my 2002 iBook still looks as good as the day I bought it.

[Vax456]

Boot speed problem? Start using an SSD and systemd

FTFY

I haven't tested my boot speeds with and without an initrd on my Gentoo system, but I can say that a kernel with OpenRC boots just as fast, if not slightly faster than kernel/initrd with Upstart.

When I tested out Exherbo, a kernel with systemd definitely boots slightly faster than any other distribution I tried out (except for distributions that specifically concentrate on boot speeds), although that's not a fair comparison since it was built from scratch and didn't have as many services running than a default Ubuntu installation.

Switching over to a kernel without an initrd will help, but not by a whole lot. A kernel with no initrd with systemd will probably help the most.

[liam]

Any chance we can get some charts looking at boot times of some of the other major distros over the past few releases?

[AnonymousCoward]

Well, I guess they should just get rid of upstart and start using systemd like everybody else.

[Lemmiwinks]

Boot speed problem? Start using an SSD and Gnome 3

[DaemonFC]

There's nothing 'broken' about expecting a filesystem to work in a sensible manner. If you can only get performance improvements by ensuring the filesystem is corrupt if the computer crashes, that's fine for a special-purpose filesystem on computers which have UPS and other systems to reduce the odds of a crash, but hopeless for general-purpose use.

And, in any case, all that wonderful performance from unsafe file writes vanishes the moment you have to tell developers 'just use fsync if you don't want your files to be corrupt after a crash'.

Up until Linux 3.1, Ext3 didn't even use write barriers.If you didn't lose data in a crash, it's only because of a set of lucky coincidences. (Think race conditions)

Hard disk are unreliable regardless of file system. Even the safest/slowest file system can become corrupt whenever there is a crash. It should be the job of the program to make sure the data it saves gets committed. It is not the filesystem's job to cripple itself and slow itself down by an order of magnitude to compensate for some barely competent application.