I am a new SlackWare convert.

[PowerMacG5]

I installed slackware 8.1 on my router today, and I must say I am impressed. SAMBA and iptables seems to be working faster than before. Do any of you think that it is worth it to kernel 2.4.19 from 2.4.18, or wait until slackware 9.0? Also, does anybody know when slackware 9.0 is coming out?

 

[n0cmonkey]

Check the kernel change log (might be illegal in the US) to see what has changed, I dont keep up on the linux kernel. Slack 9 will come out, when its finished 😉

 

[Electrode]

When I last installed Slack 8.1 on my server (a few weeks ago, damn 75GXP...) I upgraded the kernel to 2.4.19 right away. Of course, this was because I needed 2.4.19's USB support. 🙂 Also, while I'm not sure what CPU that Slack's default kernels are built for, it sure ain't the Pentium 3. 🙂

 

[riiv]

I moved my Samba server to 9.0 alpha in October and it worked finally (samba hates me). Slackware bins are still compiled to be 386 compatible.

If you''re trying to speed up your services, kill anything you don't need my Slack only runs samba and ssh.
Linux Performance Tuning

 

[Bremen]

I don't know of any pressing issues that would cause one to upgrade, however going with a 2.4.19 kernel won't hurt. Especially since you'll probably want to compile for a more recent proc (Slackware still compiles for i386). This means the minimalist nature of the distro accounts for your speed increase, as other distros now compile for i586 (I believe Debian still uses i386 though).

As for 9.0, we're all waiting man, however the old adage still applies. It'll be done when its done. :0) Grabbing -current is always possible (you can mirror any changes that happen, there shouldn't be many more before 9.0).

 

[Nothinman]

I don't know of any pressing issues that would cause one to upgrade, however going with a 2.4.19 kernel won't hurt

I would hold off, 2.4.20 RCs are out so that means it won't be long before 2.4.20 final is out.

as other distros now compile for i586 (I believe Debian still uses i386 though).

The installer kernel is for i386 so that it boots on everything, but Debian keeps precompiled kernels for almost all the different arches:

$apt-cache search kernel-image-2.4.19
kernel-image-2.4.19-386 - Linux kernel image for version 2.4.19 on 386.
kernel-image-2.4.19-586tsc - Linux kernel image for version 2.4.19 on Pentium-Classic.
kernel-image-2.4.19-686 - Linux kernel image 2.4.19 on PPro/Celeron/PII/PIII/PIV.
kernel-image-2.4.19-686-smp - Linux kernel image 2.4.19 on PPro/Celeron/PII/PIII/PIV SMP.
kernel-image-2.4.19-k6 - Linux kernel image for version 2.4.19 on AMD K6/K6-II/K6-III.
kernel-image-2.4.19-k7 - Linux kernel image for version 2.4.19 on AMD K7.
kernel-image-2.4.19-k7-smp - Linux kernel image for version 2.4.19 on AMD K7 SMP.

 

[Buddha Bart]

iptables seems to be working faster than before.

How would iptables work faster. Was your old setup honestly so overloaded that it couldn't handle network I/O?

bart

 

[Bremen]

I would hold off, 2.4.20 RCs are out so that means it won't be long before 2.4.20 final is out.

Unless I'm mistaken 2.4.20 RCs have been out and about for awhile, don't hold your breath. Look at the 2.0 and 2.2 trees, once a devel kernel tree is opened things really slow down. Of course it shouldn't take as long as the last 2.0 kernel release (couple months after the RC ;0)

The installer kernel is for i386 so that it boots on everything, but Debian keeps precompiled kernels for almost all the different arches:

My apologies, I was talking about binary files, not just the kernel. I know Debian looked into it but decided keeping seperate binary sets for each proc generation was way too much of a hassle.

 

[Nothinman]

Unless I'm mistaken 2.4.20 RCs have been out and about for awhile, don't hold your breath.

Obviously noone can tell when Marcello will decide to say it's final, but since it's been out for a while I'd say the chances of it being released soon are better. And unless there's some reason he really needs 2.4.19 there's no real necessity to upgrade anyway.

My apologies, I was talking about binary files, not just the kernel. I know Debian looked into it but decided keeping seperate binary sets for each proc generation was way too much of a hassle.

I doubt the hassle was the problem, the autobuilders already handle 11 seperate architectures, adding a few to seperate x86 wouldn't be hard. The real problem is disk space, do you have any idea who much space it takes to make a Debian mirror with ~11,000 on 11 different architectures? Now essentially make that 16 architectures if you seperate 386, 586, 686, k6 and k7. It takes a ton of disk space, and no real gain is had.