| 1 |
2009/4/2 Vladimir Skuratovich <skuratovich@×××××.com>: |
| 2 |
[...] |
| 3 |
> - For me, the startup process is slowed down considerably by mounting |
| 4 |
> a /home partition. The programs we'd want to start up as early as |
| 5 |
> possible, including X, are delayed until the mounting finishes. This |
| 6 |
> is especially noticeable with reiser4, as mounting it usually takes |
| 7 |
> a few seconds, but I suppose this is true to some extent for other |
| 8 |
> filesystems too. Maybe if the system has a /home partition we should |
| 9 |
> wait in the session manager for it to be mounted, while X is |
| 10 |
> starting up? Not sure what to do with other filesystems, such as |
| 11 |
> /usr, which still have to be mounted very early. |
| 12 |
|
| 13 |
This would need changes to the session manager, right? Is that really |
| 14 |
something that we want to be doing? |
| 15 |
|
| 16 |
[...] |
| 17 |
> - Gentoo init scripts perform a lot of 'autodetection' of different |
| 18 |
> system parameters. Probably the autodetection could be left as a |
| 19 |
> 'safe boot' option, and the real values could be recorded somewhere |
| 20 |
> and loaded at boot time? |
| 21 |
|
| 22 |
You'd need to add some quick sanity check for things like moving |
| 23 |
hard-disks across hardware. |
| 24 |
|
| 25 |
> - Instead of cleaning directories such as /var/run, they can be |
| 26 |
> mounted on tmpfs and the necessary directory tree inside can be |
| 27 |
> recreated every time. |
| 28 |
|
| 29 |
Might there be cases where these files would be needed to analyse |
| 30 |
crashes or for forensics? |
| 31 |
|
| 32 |
> - Placing static device nodes on /dev while udev is starting speeds up |
| 33 |
> the boot significantly for me |
| 34 |
|
| 35 |
By how much? |
| 36 |
|
| 37 |
> With a few hours of experiments on my system I've managed to reduce |
| 38 |
> its boot time to about ~15 seconds (not including kernel startup - and |
| 39 |
> kernel initialization takes quite long) This is on a system with a |
| 40 |
> 5400rpm laptop hard drive (40-45 mb/s linear read) with everything on |
| 41 |
> LVM inside dm-crypt. It seems that the 10 second boot time for HDD |
| 42 |
> systems, which is mentioned in Moblin presentation, is quite a |
| 43 |
> realistic goal. |
| 44 |
|
| 45 |
What kind of impact is dm-crypt having on your numbers? |
| 46 |
|
| 47 |
> If someone is interested, I can post the bootchart logs somewhere. |
| 48 |
|
| 49 |
Would be nice to look at, IMO. If you need it to be hosted somewhere, |
| 50 |
mail them to me and I'll put them up. |
| 51 |
|
| 52 |
> The system uses initng - but the structure of init scripts in Gentoo |
| 53 |
> seems very similar, so I guess porting it to Gentoo would be easy. |
| 54 |
|
| 55 |
I'm a little confused here. What system were these tests run on? |
| 56 |
|
| 57 |
Regards, |
| 58 |
-- |
| 59 |
Arun Raghavan |
| 60 |
(http://nemesis.accosted.net) |
| 61 |
v2sw5Chw4+5ln4pr6$OFck2ma4+9u8w3+1!m?l7+9GSCKi056 |
| 62 |
e6+9i4b8/9HTAen4+5g4/8APa2Xs8r1/2p5-8 hackerkey.com |
Archives