Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <679889bc-4142-41cf-8f0f-3da92a11ed3f@jeffunit.com>
Date: Mon, 7 Oct 2024 16:23:54 -0700
From: jeff <jeff@...funit.com>
To: john-users@...ts.openwall.com
Subject: Re: slow progress after --restore

On 10/7/2024 16:03, Solar Designer wrote:
> On Mon, Oct 07, 2024 at 02:59:48PM -0700, jeff wrote:
>> On 10/7/2024 14:28, Solar Designer wrote:
>>>> Based on my observations, it seems that john checkpoints somewhat
>>>> infrequently, and when it is restored,
>>>> it starts off and repeats a lot of calculations. It would be nice if
>>>> there was more fine grained check pointing
>>>> so that hashes are found much sooner. Waiting 8000 seconds before any
>>>> new hashes are found seems less than ideal.
>>>> I don't plan on restarting often, but when I do it seems to take quite a
>>>> long time to build up to full speed.
>>> That's a weird observation.  Do you mean "for about 8000 seconds" after
>>> the start of actual cracking when first per-process status lines have
>>> already appeared and timestamps have started increasing?  Or is it the
>>> loading time maybe?  Do you have a large pot file maybe?  You don't show
>>> you overriding it with --pot.
>> My pot file has about 16m entries. I mean for about 8000 seconds after
>> the process status lines start for each thread.
>>
>>> We do checkpoint often, once a minute by default (or even more often
>>> when a lot of passwords get cracked in a minute).
> Well, I can only confirm that what you observe is unexpected.  We
> checkpoint often, and moreover if you interrupt a session cleanly then
> the latest state is saved at that time.
>
> This could be a bug, or it could be something specific to your usage.
>
Is it possibly related to the --fork=22 on the command line?
Is it possible that the 'rule' starts running from the start when restarted?
I know the behavior is odd, which is why I am reporting it, and would 
like t see it fixed.
I have observed it several times.
This is for my 64 core amd epyc milan system, if that matters.

Tomorrow, (patch tuesday) I will be restarting my computer and restoring 
the john session
and I am pretty confident I will see the same issue again :-(
jeff

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.