e4rat - reduce boot time (into X) by some 50% for ext4

Blµb · 2012-02-26 17:57:28

I used to have my system boosted by using squashfs-aufs on /usr - but it only makes sense if you do not frequently update your system, and you have to be careful with kernel/udev/initscripts updates
gonna try e4rat at some point (as soon as I turn my notebook into Arch again.)

Trilby · 2012-02-28 00:22:18

I just discovered this and was skeptical. But after trying it out it is nothing short of phenomenal. It's actually a bit disturbing seeing the output of init fly by that fast as I'd never know if anything had changed. I can't wait to show a collegue how fast it boots now ... I just hope he doesn't blink!

jose1711 · 2012-03-27 08:50:04

couple of bootchart pictures from my freshly installed virtual (virtualbox hosted on windows) system: http://www.dropbox.com/gallery/29095940 … t?h=ec9501. bootchart logging stopped when my script (see sudo process) could see a window titled Firefox (max error is 1 sec as i was running sleep 1 between checks).

Unia · 2012-04-16 20:42:07

Can we run e4rat-collect again, without harming the system? I changed WM and some other applications have been added to my startup list. I'd like to "enable" e4rat for these too.

swanson · 2012-04-16 20:47:54

Yes!

EDIT; Sorry for the short answer, but I run e4rat once a month or more. It depends on the changes I make and the updates. I even had a problem booting once which was solved with a re-run. It's a great app!

Last edited by swanson (2012-04-17 05:13:13)

Unia · 2012-04-16 21:05:12

Awesome, thanks!

mariusmeyer · 2012-04-25 13:37:17

Hey guys!

Sorry if this has been posted somewhere in these 10 pages, but I couldn't find it anywhere. I'm using systemd, and I've collected files in startup.log which I've told e4rat should be under / (since my /var is on another partition). After changing to rescue.target by issuing

systemctl isolate rescue.target

I try running "e4rat-realloc /startup.log", and it prints:

Parsing file /startup.log
  17/1934 file(s) are not available
1917/1934 file(s) not on an valid ext4 filesystem

I have e4rat 0.2.2-1 and an otherwise up to date system. Issuing "mount -l" shows that /dev/sdb2 is mounted as / type ext4, which is correct. Has anyone experienced this before? Note that I've recently wiped and reinstalled Arch to use syslinux, systemd and gnome 3.4, and I didn't have this problem before (using grub and arch init)

Any help greatly appreciated!

nick_lhy · 2012-04-27 14:41:51

I have this problem too, have no idea about it ...

mariusmeyer wrote:

Hey guys!
Sorry if this has been posted somewhere in these 10 pages, but I couldn't find it anywhere. I'm using systemd, and I've collected files in startup.log which I've told e4rat should be under / (since my /var is on another partition). After changing to rescue.target by issuing
systemctl isolate rescue.target
I try running "e4rat-realloc /startup.log", and it prints:
Parsing file /startup.log
  17/1934 file(s) are not available
1917/1934 file(s) not on an valid ext4 filesystem
I have e4rat 0.2.2-1 and an otherwise up to date system. Issuing "mount -l" shows that /dev/sdb2 is mounted as / type ext4, which is correct. Has anyone experienced this before? Note that I've recently wiped and reinstalled Arch to use syslinux, systemd and gnome 3.4, and I didn't have this problem before (using grub and arch init)
Any help greatly appreciated!

kokoko3k · 2012-04-27 17:21:04

I'm tempted to try it, but i'm scared when i read comments that say that the system could not boot after e4rat, issues with sound not working or other weird things like that.
But i can't understand by myself why.

Could someone explain why it should happens, isn't e4rat "just" a sort of intelligent defragmenter?

Last edited by kokoko3k (2012-04-27 17:22:40)

swanson · 2012-04-27 18:02:52

Go ahead and try it! If, and i mean if, it possibly creates problems it is very easy to disable and if needed, uninstall.

Read instructions carefully!

SanskritFritz · 2012-04-27 20:15:27

kokoko3k wrote:

Could someone explain why it should happens, isn't e4rat "just" a sort of intelligent defragmenter?

I'm afraid there is no sane explanations, just that e4rat is not mature yet, it can contain bugs which pop up in some systems, while in others they remain hidden.

Trilby · 2012-04-27 23:28:35

kokoko3k, an "intelligent defragmenter" is only half of what e4rat is. It seems to me that e4rat-collect and e4rat-realloc could be described in that way, but e4rat-preload would not fit that description at all. The preloader starts loading files into memory before (and during) the normal init process. I also don't see where the problems would come from, but it is certainly possible that two processes running in parallel trying to access the same data could lead to some conflicts.

In otherwords, I suspect any of these alleged problems would only be with using the preload init process, not with the relocation of the files.

kokoko3k · 2012-04-28 12:48:08

Ok, just tried it and no problems so far, but with stop-watch in my hand and using e4rat-preload-lite,
here are my results to fully load a kde-desktop and firefox from the time i hit enter on grub:
1 standard: 1m:30
2 e4rat-preload-lite: 1m02
3 resume from hibernation: 18sec.

I know i asked in the past why i should choose e4rat over hibernation, but i think i'm still missing the point.

toad · 2012-04-30 09:21:30

kokoko3k wrote:

I know i asked in the past why i should choose e4rat over hibernation, but i think i'm still missing the point.

Answer is simple, you shouldn't

mariusmeyer · 2012-04-30 09:51:35

Hey guys, an update to my post above. Seems it was a bug, and it is now fixed in e4rat-git. Either wait for the next release, or modify the e4rat PKGBUILD to pull from git Here's what I did: PKGBUILD

chamber · 2012-04-30 19:01:06

Compaq Presario f500 boots to Subtle desktop with Firefox open and ready to go in 33 seconds. This took ~45 seconds previously.

solstice · 2012-05-04 10:00:52

first, you could run e4rat-realloc and never use e4rat-preload, it will still load a little faster, but not that much faster

second, I removed the FUD on the wiki page.
there was no source about the problem : when ? where ? for what ? so it' more like a FUD
about the problem with ext3 -> ext4 partition, it was caused by a bug in the 2.6.38 and 2.6.39 kernel. One walk around was to run chattr +e on each file before running e4rat-realloc. the bug is fixed in kernel 3.0rc1

As I had a ext3 -> ext4 partition, I run the chattr +e without knowing it wasn't needed.
I got not problem so far.

e4rat-preload-lite is just faster because it stops to preload only 100 files before passing to init. the library thinggy is not true anymore.

0.2.3 is out

Last edited by solstice (2012-05-04 10:04:00)

kokoko3k · 2012-05-04 19:02:36

Labello wrote:

suspending stopped working for me... which worked flawlessly on my ibm t23 until i started using e4rat. but well... since sound stopped working after resuming from suspend and it took almost as long as a cold reboot with e4rat I do not care that much.

I've had problems with standard s2disk, when going into hibernation, the system freezed jus before starting wrinting the image to disk, and i think it is caused by e4rat-preload(-lite)
Did it happened to you too?

boswbr25 · 2012-05-06 19:58:39

Hi,

I am a new user of e4rat and am loving the speed boost. Thanks to all that made this happen!

The change on my system was not as massive as others but it still reduced time to Slim then Awesome WM with Chrome & Postler email by about 20%. I have a single core Celeron 2.4GHz processor with 1GB RAM using the linux-ck kernel. It reduced startup time from 58 seconds to 47 seconds (Using a stopwatch so not 100% accurate, but definitely close).

Anyways, I had a question I was hoping someone could help with. I have Xfce as a second DE for my GF to use. If I turned the timer for e4rat-collect up to like 6000 secs and then booted to Slim and logged into Awesome, could I then log out and open Xfce to have its files stored for realloc/preload. I would then manually stop e4rat-collect.

I was just wondering if anyone had tried this. Also, if I were to boot into the standard arch kernel, does anyone know if I could still use the same e4rat startup.log and just append the e4rat-preload line to it's grub entry as well.

Thanks again to everyone developing/maintaining it. I'm wondering if there should be a section added to the Maximizing Performance wiki about this program because it seems to have "maximized my performance" more than any of the other items listed there.

Last edited by boswbr25 (2012-05-06 20:00:44)

Thme · 2012-05-26 00:20:01

I'm thoroughly impressed! I was tweaking the boot on my netbook and found a reference to this in the wiki and they weren't joking. my boot was already around 35 seconds (slim auto login and Xfce4.10 with a few background-ed daemons in rc.conf)
now the boot is between 25-27 seconds from a cold start! On an older netbook!
Specs:
1018p asus eeepc
Intel atom N450 @1.66ghz(multithreaded single core)
2gb patriot DDR2 800mhz ram installed
5400rpm seagate 250gb HDD. considering the specs it not only boots fast but also loads the majority of apps i use almost instantly in Xfce4 when combined with the preload daemon...
f#*^ing amazing!!!!

Note: recently installed a 750gb 7200rpm seagate after Cloning the 250 to it. cut down an extra 5-7 seconds on average.. I re-ran the steps with e4rat just in case. Also correction to the above numbers 27-30 was the original boot from grub as the bios load was close to 10 seconds by itself so a cold start was between 37 and 40 sec. now it's just over 30(23-24 from grub) This includes all the services and daemons I run at boot inside xfce4. still a lot faster than anything else i've ran. (ubuntu and LMDE were around 1-1 1/2 minutes). The original boot an arch before any tweaking was around 45 from a cold start.

Last edited by Thme (2012-06-24 19:37:13)

Kopfweh · 2012-05-27 18:45:33

I'm not very deep into the topic of boot-sequence, but just one question:

Does e4rat affect the init system? So is it possible to use it with both sysvinit and systemd? Or does it's usage break systemd?

dodo3773 · 2012-05-27 18:47:23

Kopfweh wrote:

I'm not very deep into the topic of boot-sequence, but just one question:
Does e4rat affect the init system? So is it possible to use it with both sysvinit and systemd? Or does it's usage break systemd?

e4rat works fine with systemd

jgreen1tc · 2012-05-27 19:30:37

I got a kernel panic around the time e4rat was preloading. I checked /var/log/boot for the errors I got but they weren't there. I'm not sure where else to look. Can someone point me in the right direction so I can research what happened?

Edit: autocorrect

Last edited by jgreen1tc (2012-05-27 19:44:33)

Unia · 2012-05-27 19:48:18

jgreen1tc wrote:

I got a kernel panic around the time e4rat was preloading. I checked /var/log/boot for the errors I got but they weren't there. I'm not sure where else to look. Can someone point me in the right direction so I can research what happened?
Edit: autocorrect

Experienced those too a couple of times, but it's gone again after a reboot. Doesn't harm anything though, at least didn't for me

swanson · 2012-05-27 19:59:46

My experience is that if you wait too long between new e4rat-collects , like a month or so, it can cause problems. After all we're on a roll.

Arch Linux

#226 2012-02-26 17:57:28

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#227 2012-02-28 00:22:18

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#228 2012-03-27 08:50:04

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#229 2012-04-16 20:42:07

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#230 2012-04-16 20:47:54

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#231 2012-04-16 21:05:12

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#232 2012-04-25 13:37:17

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#233 2012-04-27 14:41:51

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#234 2012-04-27 17:21:04

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#235 2012-04-27 18:02:52

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#236 2012-04-27 20:15:27

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#237 2012-04-27 23:28:35

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#238 2012-04-28 12:48:08

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#239 2012-04-30 09:21:30

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#240 2012-04-30 09:51:35

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#241 2012-04-30 19:01:06

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#242 2012-05-04 10:00:52

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#243 2012-05-04 19:02:36

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#244 2012-05-06 19:58:39

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#245 2012-05-26 00:20:01

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#246 2012-05-27 18:45:33

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#247 2012-05-27 18:47:23

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#248 2012-05-27 19:30:37

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#249 2012-05-27 19:48:18

Re: e4rat - reduce boot time (into X) by some 50% for ext4

#250 2012-05-27 19:59:46

Re: e4rat - reduce boot time (into X) by some 50% for ext4

Board footer