GhostBSD and OpenRC

[ASX]

kraileth, (and all concerned with boot time):
you may want to consider the purchase of an SSD disk, they start to be quite cheap, and will be considerably cheaper in the near future.

[kraileth]

kraileth, (and all concerned with boot time):
you may want to consider the purchase of an SSD disk, they start to be quite cheap, and will be considerably cheaper in the near future.

Agreed, this does speed up boot times considerably. My main laptop has an SSD for the system and a spinning hard drive for data storage and the SSD helps a lot. Still Linux is of course booting up much, much quicker. However I'd also say that the FreeBSD kernel is taking much more time to do it's thing before the actual init process kicks in. And neither OpenRC nor Runit or any other project can be of any help here. Probably the time that bsd-init takes is not all that bad actually. Still it would be helpful if it could be reduced (which would be possible with a different init)!

If I had the money, I'd love to try out an NVMe. But those won't be available at affordable prices anytime soon, I guess.

[ASX]

However I'd also say that the FreeBSD kernel is taking much more time to do it's thing before the actual init process kicks in. And neither OpenRC nor Runit or any other project can be of any help here. Probably the time that bsd-init takes is not all that bad actually. Still it would be helpful if it could be reduced (which would be possible with a different init)!

Good point!
Grub2 is a bit faster than BTX at loading kernel, yet the kernel initialization phase remain the same.

If I had the money, I'd love to try out an NVMe. But those won't be available at affordable prices anytime soon, I guess.

Add me too, unfortunately that is going to require a new machine.

[NevilleGoddard]

Hi, kraileth.
Yes, I'd be happy to tinker with it some more and I'll write something up. I don't know how successful I'll be. I'll try and get it done before the weekend. I'll also look into Runit. I had a look at a Kris Moore post and he says he sped up boot times on TrueOS from one minute to 10 seconds if this can be done on GhostBSD it would pretty much solve your problem.

[ericbsd]

Not that I am against of booting faster, but why booting faster would solve a problem?

The boot time is not a real problem or issues. Believe me or not on my ssd i do not se much difference from linux and GhostBSD with Grub install.

I have been using FreeBSD for since 2008 my problem has never been boot time can past a month or more without rebooting. But I understand that some people need that because they close their PC when they are not on it boot time is better for those people but how much 20 second would matter.

[NevilleGoddard]

I was referring to kraileth's slow boot problem. He said he's pretty much in need of a faster boot time.

[ericbsd]

There is so much init openlaunchd, Jobd, runit, openrc and more, why should we go with OpenRC or runit and not Jobd. Not only that changing the init would mean to rewrite tons of ports, that are made to run the current init and NetworkMgr would not be compatible with FreeBSD after that unless I code it to run on all init possible.

There would be to much broken ports, I did not think of all the work of an init would take until now. GhostBSD system only have part time development and it is mainly me for now. ASX is taking care of lot extra work like helping here, debugging and he is working on our future repository, convbsd we do not have any news from him yet I hope he is still well but he was helping developing on the system side, kraileth he is fresh new I do not know what would be is main focus in future, but even if get on the system dev side that is not enough to maintain ports and system to wirk with a new init.


Like I always say in the past if people want to see thing happen step in the project and make things happen, until then thing will be slow, and now excuse of I do not know how to program, I have not finish high school and did not know anything about programming, Linux, BSD, build system and anything like this and today I am the founder this project that. If someone would had told me 10 year ago that I would do a custom FreeBSD system, I would had say what is FreeBSD at that time the only thing I knew about PC is installing game on windows XP and recording music.

[ericbsd]

NevilleGoddard dont take that toward you, right now I am mostly documenting My thought and in the same time making point that wee need people to make things happen.

[NevilleGoddard]

Thanks for the reply, ericbsd and I understand. If I can help GhostBSD in any way, please let me know.

[ASX]

if people want to see thing happen step in the project and make things happen

Yep, then they will start to realize that some of their ideas it is not so great when they achieve some result at the price to broke a few different things.

One thing is looking at a single improvement, another thing is looking at the same improvement while considering the complete project and the related dependencies.

Specifically about startup services concurrently:
- GhostBSD by default enable very few services
- when a user is going to add lot of services and startup them in parallel (where possible) the system is going to have a lot of disk I/O contention because of simultaneous requests ... that cannot be served as required from a conventional disk, in other word the disk will become the bottleneck, which means parallel startup will be faster only when using SSDs or high speed disks.

I recently recovered an old Dell laptop (Latitude d430, circa 2008) and substituted the original 1.8" IDE disk with a cheap 32 GB msata SSD, after that the boot time up to a full XFCE desktop is less than 30 seconds.
Opening a terminal an launching "uptime" it shows 21 or 22 seconds.

Anyone here will have a very hard time to convince me to switch init system to gain a few more seconds.