Re: [squid-users] (Maybe) Bugreport: Ext3-fs error with squid 2.4stable6

Marc Elsen wrote:
>
> Simon White wrote:
>
>>11-Apr-02 at 13:42, Henrik Nordstrom (hno@marasystems.com) wrote :
>>
>>>Any errors reported by the filesystem in your syslog/messages is by
>>>definition kernel or hardware problems.
>>>
>>>The fact that you don't receive these using 2.4.STABLE4 is purely
>>>coincidental.
>>
>>Not so much coincidence, as that /perhaps/ there is code in STABLE6 which
>>causes code within the kernel disk I/O routines to screw up. However it is
>>clear that Squid will just make read/write demands to the hardware, and it
>>is the kernel's job to do this in conjunction with filesystem drivers.
>>
>>ext3fs is still reasonably new and was marked as EXPERIMENTAL on my last
>>kernel compile some weeks ago.
>>
>>I think using another filesystem with Squid STABLE6 would prove this one
>>anyway. I would suspect kernel code / SCSI driver / combinations of
>>these...
>>
>
>
> Hmm, I believe that the original poster's problem
> should >at least< be verified on another system with EXT3
> before SQUID is blamed.
> I mean more exactly : before it can be claimed that SQUID
> would have tricked a bug in EXT3; which has an app. can never
> be excluded, that is true.
> Programs have been know to crash os-es too...

If you have an application that can break an OS, without malicious
tricks (fork-bomb, malloc explosion, etc.) then you still have a broken
/OS/. It may be that the app is behaving incorrectly--but it should
never be able to break the OS if it resides in userspace.

-- 
Joe Cooper <joe@swelltech.com>
http://www.swelltech.com
Web Caching Appliances and Support