Because your MySQL database needs to write its state to the disk. Considering this is taking longer than 5 minutes, you probably have a very slow disk. A hardware upgrade may be in order if you plan to run a database at this scale.
Generally, yeah, they send a sigterm first but if the program doesn't respond in a few seconds, it gets blown away. it had its chance to shut down in an orderly fashion. As far as I'm aware only systemd is so incredibly deferential as to give things 90s, 10m, or indefinite (!) timeouts.
systemd is RHEL project, and rhel is basically server os, so giving big timeouts makes sense
on desktop, nobody needs to have large timeouts, 30 seconds and not down, fricking SIGKILL
What if your DB takes longer than 5 seconds to commit its data? Do these inits just blindly cause corruption and inconsistency?
systemd is RHEL project, and rhel is basically server os, so giving big timeouts makes sense
on desktop, nobody needs to have large timeouts, 30 seconds and not down, fricking SIGKILL
So these alternative inits aren't viable for server usage?
2 years ago
Anonymous
>So these alternative inits aren't viable for server usage?
the kill timeout is (or at least it should be unless you use shit init) configurable somewhere, at least globally and in some cases per-service
2 years ago
Anonymous
Isn't that also the case for systemd?
People only hate systemd because it is not "modular", aka it's one large program and not 10 different programs and libraries, and if you ask me, that's based
2 years ago
Anonymous
>corruption and inconsistency
dbms should already have means to prevent them in any case, like writing to a separate file and then renaming it, overwriting the original
you can change the time limits, i turned mine down to 30s since i don't use anything that should take that long anyway
if you find yourself reaching the limit regularly, then you should fix your shit, either increase the limit so your shit can finish, or change your shit so it doesn't take as long to finish
this is actually not a systemd problem
Because your MySQL server either has a shitload cached in RAM that needs to be written back, or you've configured it shittily. And unless you have ECC RAM and a UPS, allowing it to cache that much without writing it back after a reasonable timeout is configuring it shittily.
MacOS sends a message to home in Cupertino every time you launch a program.
And if the server is down, you can't open any programs.
Nice proprietary corporate surveillance garbage you have there....
>disconnect computer from internet >can't launch any programs
Sounds fake... holy shit I just tried it and it's true. Why aren't more people talking about this??
>"muh stop job!!! wtf!!!!" >the daemon is a defective piece of shit
every time
i don't care much for systemd but it's singlehandedly forced a shit ton of flaky projects to finally fix their shit instead of forcing distros to package their own ad hoc kludges
Because your MySQL database needs to write its state to the disk. Considering this is taking longer than 5 minutes, you probably have a very slow disk. A hardware upgrade may be in order if you plan to run a database at this scale.
that's not linux, that's systemd
it tries to safely close the programs
i got annoyed by that and installed a linux distro without systemd
Do other inits really just send programs a SIGKILL even when they're in the middle of doing something?
this is often the case yes
Nobody would use such an init in production.
This causes things to become unstable and break. I remember my distro dying within a month. Ever since systemd it's been incredibly stable.
Generally, yeah, they send a sigterm first but if the program doesn't respond in a few seconds, it gets blown away. it had its chance to shut down in an orderly fashion. As far as I'm aware only systemd is so incredibly deferential as to give things 90s, 10m, or indefinite (!) timeouts.
systemd is RHEL project, and rhel is basically server os, so giving big timeouts makes sense
on desktop, nobody needs to have large timeouts, 30 seconds and not down, fricking SIGKILL
What if your DB takes longer than 5 seconds to commit its data? Do these inits just blindly cause corruption and inconsistency?
So these alternative inits aren't viable for server usage?
>So these alternative inits aren't viable for server usage?
the kill timeout is (or at least it should be unless you use shit init) configurable somewhere, at least globally and in some cases per-service
Isn't that also the case for systemd?
People only hate systemd because it is not "modular", aka it's one large program and not 10 different programs and libraries, and if you ask me, that's based
>corruption and inconsistency
dbms should already have means to prevent them in any case, like writing to a separate file and then renaming it, overwriting the original
runit waits 1 sec
Could also lower the time limit
you can change the time limits, i turned mine down to 30s since i don't use anything that should take that long anyway
if you find yourself reaching the limit regularly, then you should fix your shit, either increase the limit so your shit can finish, or change your shit so it doesn't take as long to finish
this is actually not a systemd problem
youd think someone running a mysql server wouldnt be this stupid
Only people running mysql garbage are complete morons
Anyone willingly using MySQL instead of MariaDB nowadays knowing that is owned by Oracle aren't really the brightest of the crowd.
Millions of lines of garbage bloat
Because your MySQL server either has a shitload cached in RAM that needs to be written back, or you've configured it shittily. And unless you have ECC RAM and a UPS, allowing it to cache that much without writing it back after a reasonable timeout is configuring it shittily.
dude just unplug it like lmao
That's what I do to my machine that I use on my TV, hasn't done any harm yet.
Well it takes time to send all the hashes from your pc to NSA for backup purposes. You don't have nothing to hide anyway... right?
macOS doesnt have these freetroon problems
MacOS sends a message to home in Cupertino every time you launch a program.
And if the server is down, you can't open any programs.
Nice proprietary corporate surveillance garbage you have there....
>disconnect computer from internet
>can't launch any programs
Sounds fake... holy shit I just tried it and it's true. Why aren't more people talking about this??
You installed it incorrectly or you're about to lose a lot of data. Good job OP.
Also this is not a tech support board. Ask
next time homosexual
good job falling for low effort bait, homosexual
they need to finish sending your data to someone in internet doing surveillance, before closing it down
this is why it varies widely on the very same computer, how long the shutdown sequence takes each day
At least it's not that Black person watchdog
>"muh stop job!!! wtf!!!!"
>the daemon is a defective piece of shit
every time
i don't care much for systemd but it's singlehandedly forced a shit ton of flaky projects to finally fix their shit instead of forcing distros to package their own ad hoc kludges
I simply cut the power when I am done using my Raspberry Pi and I haven't had any issues
Is artix as a sever os viable? You can probably configure the init to give a process a maximum timout of 30 seconds and then blow it out.