On Sun, Aug 25, 2024 at 11:55 PM Tim via users
<users@xxxxxxxxxxxxxxxxxxxxxxx> wrote:
>
> Fulko Hew:
> > > But 'shutdown' provided all the housekeeping work such as:
> > > - disabling logins
> > > - sending out messages to users screen warning them of the impending
> > > doom
> > > - providing grace time
> > > - unmounted file systems
> > > - killed the system
>
> Patrick O'Callaghan:
> > Same here. However these are reasonable measures on a multi-user
> > system. On a single-user desktop they just get in the way, especially
> > with journal-based filesystems.
>
> Things really oughta order unmounts, unmounts should happen pretty
> quickly, be flagged as done, and the progress of shutting down be
> monitored. Not, order unmounts, wait some time, and assume it worked.
> Likewise for various other shutdowns.
>
> Things may have terminated almost instantly without issues, things may
> have jammed and it still wouldn't have waited long enough. If
> something sticks you really ought to be prompted about it. If you
> don't have network mounts, databases running, mail servers currently
> dealing with a queue, etc, I see no excuses for prolonged shutdowns.
I think part of the problem is the behavior of sync(8) and sync(2)
(which the sync command uses) are poorly specified.[1,2] The man pages
don't even clearly state whether the command or api call block or
return immediately. If it blocks until complete, then the sync family
are synchronous; if they return immediately they are asynchronous. If
the sync family is asynchronous (returns immediately), then a script
will need to insert sleeps in hopes there's enough time for the
write-backs to get done.
The sync(2) man page does say this:[2]
According to the standard specification (e.g., POSIX.1-2001), sync()
schedules the writes, but may return before the actual writing is done.
However, since version 1.3.20 Linux does actually wait. (This still
does not guarantee data integrity: modern disks have large caches.)
But that leaves additional questions since the wait or blocking
appears to be an incomplete wait.
[1] https://linux.die.net/man/8/sync
[2] https://linux.die.net/man/2/sync
Jeff
--
_______________________________________________
users mailing list -- users@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to users-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/users@xxxxxxxxxxxxxxxxxxxxxxx
Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
