SlackwareThis Forum is for the discussion of Slackware Linux.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
Distribution: Slackware64 15.0 (started with 13.37). Testing -current in a spare partition.
Posts: 928
Rep:
90 seconds to boot? Are you counting from when you press the power button or is this 90 seconds after Lilo?
Slackware boots in 17 seconds here with a few things in rc.local (the 'compat' option in Lilo is really magic).
Post takes around 6 seconds, plus 5 seconds in Lilo screen.
The last time I looked at Slackware boot time, a big part of it was waiting on DHCP timeouts on ethernet network interfaces which might or might not even be plugged in (defaulting to 30 seconds per interface, so a minute on a dual-ethernet system).
Detecting carrier (whether the NIC is plugged in) from software is supported for -some- hardware, but not other hardware, and in the latter case all you can do is go through the motions on the assumption that there is a network attached, and deal with the DHCP timeout.
A better (albeit incomplete) solution would be to try to detect carrier on systems which support that, and skip DHCP entirely on interfaces not attached to a network, and fall back to the current behavior on systems which do not support such detection.
I looked at implementing that, but realized that I simply don't reboot my systems frequently enough to make it worthwhile, and put the idea on permanent back-burner.
I did reduce my DHCP timeout to 8 seconds in rc.inet1.conf though (DHCP_TIMEOUT). On my networks if DHCP doesn't happen in less than 8 seconds, it's never going to happen. This is not a solution suitable for every network.
If all of the interfaces defined in rc.inet1.conf are either assigned static addresses or plugged into a network with fast DHCP, of course, these delays do not manifest.
I have a Samsung SSD and Slackware is fast - and I don't even have TRIM (noatime) enabled in fstab, I am using JFS. I still hope later down the line maybe Slackware will allow installation using a FS based on NAND devices (JFFS2, YAFFS, F2FS). Don't get me wrong, the more 'traditional' fs work, but again with the ever growing adoption of solid state NAND devices, I still feel that using a FS that is solely optimized for such drives would be a logical idea.
Just as an FYI, trim and noatime are two different things. Trim is enabled in the fstab using the discard option (or using fstrim manually through the CLI). noatime, while typically suggested for SSDs, prevents updating access times for files and folders when they're accessed. It is commonly used on SSDs to minimize the writes to the NAND (although people would enable them on HDDs to minimize extra writes that can cause thrashing), but it probably wouldn't be enough to cause the NAND to fail for too many writes. That being said, I think I enabled it on my SSD (I'm at work and too lazy to ssh into my computer to check), mainly because I don't care if access times are recorded.
Just as an FYI, trim and noatime are two different things. Trim is enabled in the fstab using the discard option (or using fstrim manually through the CLI). noatime, while typically suggested for SSDs, prevents updating access times for files and folders when they're accessed. It is commonly used on SSDs to minimize the writes to the NAND (although people would enable them on HDDs to minimize extra writes that can cause thrashing), but it probably wouldn't be enough to cause the NAND to fail for too many writes. That being said, I think I enabled it on my SSD (I'm at work and too lazy to ssh into my computer to check), mainly because I don't care if access times are recorded.
Distribution: LFS 9.0 Custom, Merged Usr, Linux 4.19.x
Posts: 616
Rep:
As with most things in Linux distributions, configuration has huge effects on speed, disk usage, battery life, etc. From startup times to app load times, the configuration will have far more effect on these things than whether you're using Systemd, SystemV, or whatever.
Slackware has always been slow with the default fattykernel.
In addition to what's already been said, booting Slackware in EFI mode will also be faster.
How do you set that up? By having the installation medium botted in EFI mode.
I always wanted to mess with EFI on Slackware (on Virtualbox) - but the latest Virtualbox release still doesn't work with EFI. It no longer crashes this time when I try EFI, but the bootup is stuck on GRUB and is frozen when trying to boot the Slackware ISO.
90 seconds to boot? Are you counting from when you press the power button or is this 90 seconds after Lilo?
Slackware boots in 17 seconds here with a few things in rc.local (the 'compat' option in Lilo is really magic).
Post takes around 6 seconds, plus 5 seconds in Lilo screen.
Yes. From the time I press the power button to tty1 user login terminal.
Also, I did not count the time that LILO counts down 2 minutes for the boot menu.
The last time I looked at Slackware boot time, a big part of it was waiting on DHCP timeouts on ethernet network interfaces which might or might not even be plugged in (defaulting to 30 seconds per interface, so a minute on a dual-ethernet system).
Detecting carrier (whether the NIC is plugged in) from software is supported for -some- hardware, but not other hardware, and in the latter case all you can do is go through the motions on the assumption that there is a network attached, and deal with the DHCP timeout.
A better (albeit incomplete) solution would be to try to detect carrier on systems which support that, and skip DHCP entirely on interfaces not attached to a network, and fall back to the current behavior on systems which do not support such detection.
I looked at implementing that, but realized that I simply don't reboot my systems frequently enough to make it worthwhile, and put the idea on permanent back-burner.
I did reduce my DHCP timeout to 8 seconds in rc.inet1.conf though (DHCP_TIMEOUT). On my networks if DHCP doesn't happen in less than 8 seconds, it's never going to happen. This is not a solution suitable for every network.
If all of the interfaces defined in rc.inet1.conf are either assigned static addresses or plugged into a network with fast DHCP, of course, these delays do not manifest.
I tried "chmod -x rc.inet1" and the boot time did reduce. However, I have to manually turn on the DHCP services later on after booting into the screen.
And what does it mean by "try to detect carrier on systems which support that"?
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.