[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#838503: debian-installer: mdadm should not start syncing RAID1 arrays at full speed during installation

[ Responding to this old bug, prompted by discussion in #debian-boot
  tonight ]

On Sat, Jun 24, 2017 at 09:01:49PM +0200, Cyril Brulebois wrote:
>Hi Baptiste,
>Baptiste Jonglez <debian@bitsofnetworks.org> (2016-09-21):
>> When creating a RAID1 array in the debian-installer and using it for
>> the installation, mdadm immediately starts syncing the disks of the
>> RAID array.
>> This is a bad idea, because the subsequent install will be really slow
>> on rotational disks (linear disk access by mdadm and random disk
>> access by dpkg).  On a fairly recent computer with 2 SATA disks, the
>> installation took around 20 minutes before even arriving to the
>> tasksel step.
>Well, I can understand the argument, but the user could very well expect
>disks to be in mirror mode as soon as possible, and not wait until a few
>days after the installation before the initial sync ends…

The number isn't persistent across reboots, so it would speed up again
on the reboot after the installation finished.

>> I can see two solutions:
>> 1) lower the speed of the syncing operation, by setting the
>>    "dev.raid.speed_limit_max" sysctl setting to e.g. 1000;
>I'm not sure it would be a good idea to stick a hardcoded value there,
>even if we were to lower the available bandwidth… Too many different
>cases, be it about actual hardware, disk sizes, etc.

1000 is a nice small number, which I think might make sense as a
default *while* we're installing. Once the installer has finished and
the system is rebooted, all will go back to normal. Hell, we could
even expose this value via debconf for expert (and preseed) if needed.

Steve McIntyre, Cambridge, UK.                                steve@einval.com
"Yes, of course duct tape works in a near-vacuum. Duct tape works
 anywhere. Duct tape is magic and should be worshipped."
   -― Andy Weir, "The Martian"

Reply to: