Re: [Tails-dev] Please review & merge bugfix/less-aggressive…

Delete this message

Reply to this message
Autore: intrigeri
Data:  
To: The Tails public development discussion list
Oggetto: Re: [Tails-dev] Please review & merge bugfix/less-aggressive-hard-disk-APM-on-AC
Hi,

Maxim Kammerer wrote (02 Mar 2013 20:39:35 GMT) :
> On Sat, Mar 2, 2013 at 9:49 PM, intrigeri <intrigeri@???> wrote:
>> If these settings were seriously
>> wrong, I guess we would have seen quite a lot of bug reports in Debian
>> during the Wheezy development cycle.


> Like these? Just a few examples.
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=684241
> https://bugs.launchpad.net/ubuntu/+source/hdparm/+bug/913050


Maxim: thanks for the pointers. I'm now convinced (and have started
doing what's needed to fix this in Wheezy, by the way).

Tails developers: I've rewritten the branch history to use 128 instead
of 127, and improved the commit message to explain why.

>>> Having 254 for everyone on AC is unreasonable, in my opinion.
>>
>> I concur with Alan: would you please share some insight about why this
>> is unreasonable?


> I don't think hard drives will ever spin down with -B254, even with -S
> (again, it has been some time, so I might be wrong — I think the
> laptop on which I saw problems with -B127 never spun down with -B254,
> so this might not be universally true/untrue). Given that the people
> recommending use of -B254 don't really understand the issue (e.g.,
> -B128 would fix the issues with a drive constantly going idle from due
> to -B127 just the same, and sometimes the problems result from using
> -B255 with older hdparm versions that didn't treat the parameter
> correctly), I would say that -B128 is the correct value to use, when
> -S is also specified (which laptop-mode does for IDLE_TIMEOUT
> parameters).


Thanks for the explanation.

There's
https://ata.wiki.kernel.org/index.php/Known_issues#Drives_which_perform_frequent_head_unloads_under_Linux,
but after a quick look at a few reports, they're indeed either old or
not totally convincing (i.e. there's no indication that -B 128 was not
enough), so at this point I could be convinced -B 128 is the way to go.

However: the original bug we had reported against Tails
0.16 a few weeks ago made it clear that the default -B 128 triggered
the problem them, while -B 254 fixed it. Even if we assume that most
cases were reported and analyzed by people who "don't really
understand the issue", and would simply be solved by -B 128, there's
apparently at least one case in the wild where -B 254 is needed.
I believe that one diagnosed, debugged and reported case against Tails
means many more potential real world cases. So, I'm still in favour of
using -B 254 for Tails.

But well, I guess Maxim will come out with a few bug reports that
clearly show that -B 254 is harmful ;)

If this does not happen within a few days, my merge request holds.

Cheers,
--
intrigeri
| GnuPG key @ https://gaffer.ptitcanardnoir.org/intrigeri/intrigeri.asc
| OTR fingerprint @ https://gaffer.ptitcanardnoir.org/intrigeri/otr.asc