4 ms·

I hear people always complaining about 'needless' changes like systemd over init. But I sure like how my modern database servers reboot in less than 30 seconds

I hear people always complaining about 'needless' changes like systemd over init. But I sure like how my modern database servers reboot in less than 30 seconds, vs 5-12 minutes when they were running RHEL 5. (yes, some of that is because we moved from BIOS to UEFI)

zinekeller 5y ago
Hmm, why do I feel that a) the older servers verifies memory banks fully before boot (which takes up minutes) and b) the change to solid-state media has the biggest impact and not on SystemD. Can you confirm that at least a) is not the reason for slow boot times?
- antoinealb 5y ago
  I know that anecdote is not data, but I had an Arch Linux laptop when the distribution moved from Init scripts (SysV init ?) to systemd, and the boot time was easily cut in three, going from 30s to 10s, on exactly the same laptop. Of course switching from HDD to SSD was a huge improvement, but don't discount what systemd's efficient boot parallelism was able to achieve.
  - trasz 5y ago
    It’s worth keeping in mind that the old Linux rc scripts where quite mediocre compared to their BSD counterparts. They didn’t even have a dependency mechanism. So, it’s fine to compare sysv Linux scripts to systemd, but please don’t extrapolate that to other systems.
    - antoinealb 5y ago
      I did not know that the BSD ones are more advanced, thanks for pointing it out. Out of curiosity, are there alternative init systems in the BSD world ?
      - trasz 5y ago
        Sure, eg OpenRC.
lordgroff 5y ago
Sure (although 5-12 minutes is very odd no matter how you slice it), but that's a bit of a false choice. I have nothing against systemd personally, but I recognize more and more that a good part of the reason is because I never had to use it in anger. There were other choices that Linux could have taken rather than sysvinit vs init.d and some distros through the years have taken different approaches. FreeBSD's rc.d is not sysvinit either.
- throwawayboise 5y ago
  The Linux machines that I have managed that took anywhere close to that long to boot (older IBM and HP) just have a very lengthy BIOS/POST start-up process. Once you got through that, the OS (RHEL 5 at the time) booted up pretty quickly.
- throwawaysysd 5y ago
  >There were other choices that Linux could have taken rather than sysvinit vs init.d I don't think that is true. If you look at those other distros you'll mostly find that it either didn't pan out, or it's quickly approaching the same design and architecture of systemd, where distros start using declarative prgramming to integrate tooling around common workflows. This is inevitable when you consider the constraints on the system as a whole. Other related service management tools like Docker are under the same constraints and you'll notice those have a similar architecture too.
mnd999 5y ago
I wouldn’t consider reboot time to be a massive selling point for a database server though. Surely you don’t want to be doing that very often?
- regularfry 5y ago
  On the other hand, when you do, you want it back up now.
  - 1500100900 5y ago
    If it's something like a galera cluster then not really, I think. You want the updated machine to become operational again in a reasonable amount of time, because during the time it's down you have fewer points of failure. But the difference between 2 minutes and 5 minutes in this case is not a big deal, in my opinion.
  - mnd999 5y ago
    Yeah, maybe you're right. I actually had this conversation with my PM the other day - I work on a fairly well known database. They were asking us to improve the startup time of the product.
  - kazen44 5y ago
    actually, i want it backup up concurrently. for instance, the systems better checks all memory banks and disks before it finishes booting. Having a system boot slower is an annoynance when downtime occurs, but having a system boot incorrectly is far worse.
brimble 5y ago
> 5-12 minutes I've run a lot of Linux machines on a lot of hardware for a lot of years and have no idea what could have been going on that would cause a 5+ minute boot, that was also unnecessary enough that a different BIOS could get it down to 30s. Back in the bad old days of circa ~2000 when I ran Linux on a variety of garage-sale potatoes, I don't think any took 5 minutes to boot.

zinekeller 5y ago

Hmm, why do I feel that a) the older servers verifies memory banks fully before boot (which takes up minutes) and b) the change to solid-state media has the biggest impact and not on SystemD. Can you confirm that at least a) is not the reason for slow boot times?

antoinealb 5y ago
I know that anecdote is not data, but I had an Arch Linux laptop when the distribution moved from Init scripts (SysV init ?) to systemd, and the boot time was easily cut in three, going from 30s to 10s, on exactly the same laptop. Of course switching from HDD to SSD was a huge improvement, but don't discount what systemd's efficient boot parallelism was able to achieve.
- trasz 5y ago
  It’s worth keeping in mind that the old Linux rc scripts where quite mediocre compared to their BSD counterparts. They didn’t even have a dependency mechanism. So, it’s fine to compare sysv Linux scripts to systemd, but please don’t extrapolate that to other systems.
  - antoinealb 5y ago
    I did not know that the BSD ones are more advanced, thanks for pointing it out. Out of curiosity, are there alternative init systems in the BSD world ?
    - trasz 5y ago
      Sure, eg OpenRC.

lordgroff 5y ago

Sure (although 5-12 minutes is very odd no matter how you slice it), but that's a bit of a false choice. I have nothing against systemd personally, but I recognize more and more that a good part of the reason is because I never had to use it in anger. There were other choices that Linux could have taken rather than sysvinit vs init.d and some distros through the years have taken different approaches. FreeBSD's rc.d is not sysvinit either.

throwawayboise 5y ago
The Linux machines that I have managed that took anywhere close to that long to boot (older IBM and HP) just have a very lengthy BIOS/POST start-up process. Once you got through that, the OS (RHEL 5 at the time) booted up pretty quickly.
throwawaysysd 5y ago
>There were other choices that Linux could have taken rather than sysvinit vs init.d I don't think that is true. If you look at those other distros you'll mostly find that it either didn't pan out, or it's quickly approaching the same design and architecture of systemd, where distros start using declarative prgramming to integrate tooling around common workflows. This is inevitable when you consider the constraints on the system as a whole. Other related service management tools like Docker are under the same constraints and you'll notice those have a similar architecture too.

mnd999 5y ago

I wouldn’t consider reboot time to be a massive selling point for a database server though. Surely you don’t want to be doing that very often?

regularfry 5y ago
On the other hand, when you do, you want it back up now.
- 1500100900 5y ago
  If it's something like a galera cluster then not really, I think. You want the updated machine to become operational again in a reasonable amount of time, because during the time it's down you have fewer points of failure. But the difference between 2 minutes and 5 minutes in this case is not a big deal, in my opinion.
- mnd999 5y ago
  Yeah, maybe you're right. I actually had this conversation with my PM the other day - I work on a fairly well known database. They were asking us to improve the startup time of the product.
- kazen44 5y ago
  actually, i want it backup up concurrently. for instance, the systems better checks all memory banks and disks before it finishes booting. Having a system boot slower is an annoynance when downtime occurs, but having a system boot incorrectly is far worse.

brimble 5y ago

> 5-12 minutes I've run a lot of Linux machines on a lot of hardware for a lot of years and have no idea what could have been going on that would cause a 5+ minute boot, that was also unnecessary enough that a different BIOS could get it down to 30s. Back in the bad old days of circa ~2000 when I ran Linux on a variety of garage-sale potatoes, I don't think any took 5 minutes to boot.