Re: [RFC] obsoleting cache_effective_group from 3.2

2009-02-11 Thread Kinkie
On Wed, Feb 11, 2009 at 2:56 AM, Amos Jeffries squ...@treenet.co.nz wrote: I'm opening this old discussion up again. [...] HISTORY: If I recall correctly, the only holdback we had last time this was discussed was that certain setups and winbind needed it to work. That has since changed

Re: [RFC] obsoleting cache_effective_group from 3.2

2009-02-11 Thread Henrik Nordstrom
ons 2009-02-11 klockan 14:56 +1300 skrev Amos Jeffries: WHY: * it's a security breach. Why? * it's the source of many permissions annoyances. Yes. * the setting is still widely recommended in online how-to's Yes, and often for the wrong reasons. * current Squid-3+ are perfectly

Re: [RFC] obsoleting cache_effective_group from 3.2

2009-02-11 Thread Amos Jeffries
ons 2009-02-11 klockan 14:56 +1300 skrev Amos Jeffries: WHY: * it's a security breach. Why? Overriding the underlying OS, which admin may understand, with behavior they may not. Can cause them to enact less secure workarounds; I have seen squid effective-user'd to the root UID not long

Re: [RFC] obsoleting cache_effective_group from 3.2

2009-02-11 Thread Henrik Nordstrom
tor 2009-02-12 klockan 12:30 +1300 skrev Amos Jeffries: Overriding the underlying OS, which admin may understand, with behavior they may not. Can cause them to enact less secure workarounds; I have seen squid effective-user'd to the root UID not long ago. cache_effective_user root is not

[RFC] obsoleting cache_effective_group from 3.2

2009-02-10 Thread Amos Jeffries
I'm opening this old discussion up again. WHY: * it's a security breach. * it's the source of many permissions annoyances. * the setting is still widely recommended in online how-to's without reference to the security problems playing with it causes. * current Squid-3+ are perfectly capable