How does readahead(2) affects a iscsi device

[Maddin]

Hi folks,

Sorry but not so firm in kernel hacking and structures, but if I've
understood this correctly it should perform normally?!
The problem is that I do a readahead (the tool) on a iscsi device with
infortrend iscsi san as backend to warm up caches and my connection died,
correctly the whole controller died and have to be reseted.

Has anyone an idea?

Cheers
Maddin

[Ulrich Windl]

On improperly mastered (?) DVDs there once (years ago) was a problem that read-
ahead would exceed the end of the device, causing errors. I don't know for iSCSI,
but generally reading infrequent large chunks is better than reading frequent
short junks.

Regards,
Ulrich


>
> Cheers
> Maddin
>
>
>
>
>
> >

[Mike Christie]

It should normally work. The iscsi layer is basically just passing
around scsi commands, so it does not know if read ahead is being used or
now and should not care. I do not know about specific targets and the
settings they need.

[Konrad Rzeszutek]

Whoa! Have you filled a ticket with them?
>
> Has anyone an idea?

Sounds like a big bug in their product.
>
> Cheers
> Maddin
>
>
>
>
>
>

[Mailingliste]

Hello Konrad,

> On Fri, Jul 03, 2009 at 04:10:19PM +0200, Maddin wrote:
>
>> Hi folks,
>>
>> Sorry but not so firm in kernel hacking and structures, but if I've
>> understood this correctly it should perform normally?!
>> The problem is that I do a readahead (the tool) on a iscsi device with
>> infortrend iscsi san as backend to warm up caches and my connection died,
>> correctly the whole controller died and have to be reseted.
>>
>
> Whoa! Have you filled a ticket with them?
>

Yes sir, ticket is opened. But unfortunately they can't reproduce the
whole bahvior. At their tests, the nic was stopping answering on pings,
and now this ticket was escalated to 2nd level support. Now i have to wait.