Question about iSCSI / NAS / Cluster

[Francisco Cabrita]

Hi Bob,

Thanks for your reply :)


On Mon, May 25, 2009 at 5:25 PM, Bob McClure <h8msft at gmail.com> wrote:

> Francisco,
>   This sounds very similar to the issue I encountered when trying to
> backend a shared SAN connection. The problem that
> I ran into was that this shared area would need to be configured as a
> shared file system such as GFS that is not available
> for FreeBSD. I did get concurrent connections, but bad things happened.


Yep and those bad things did happen...(love backups)
Yes that's sad we don't have such GFS but maybe CodaFS will be ready to
production one day.
(I have no idea how about MogileFS is regarding distributed FS under
FreeBSD)


>
>
>   There is package that is available for FreeBSD (ifstated) that reportedly
> will aid you with knowing when a carp status changed, but
> I personally never got it to work properly due to time constraints.


I know and I use ifstated to switch from Cable to ADSL Internet providers.
It's a KISS solution. I didn't even think using it on this scenario... :)
gonna think about it.


>
>   My original solution that I still use is a script that looks at the carp
> device info from ifconfig and reports MASTER or BACKUP.
> This can be executed periodically by a script to have the new master mount
> the shared resource and the non-master give it up.
> This is far from ideal, but works in instances where immediate fail over is
> not required. This is designed to be called from a shell
> script via backtick.


I see. link up for the master; test connection and in case of some network
failure, get that target offline and mount another to slave one.


>
>
> STATUS=`./carpstatus.pl`
> if [ $STATUS = "MASTER" ] ; then
> .....
> fi
>
> (carpstatus.pl)
> #!/usr/bin/perl
> # $output = `ifconfig | grep carp: | grep vhid | /usr/bin/awk '{print
> \$2}'` ;
> # chomp($output) ;
> # print "$output" ;
>
>   I hope this helps you with your issue.


Carp sound really good to this. I was reading about Lagg operating as a
failover but that don't make sense because I want to see and test both links
but use just one at a given time.

Very cool, and thanks very much for your tip! :)

Best regards,
Francisco Cabrita



>
> Bob McClure
>
>
>
> On Mon, May 25, 2009 at 9:39 AM, Francisco Cabrita <
> francisco.cabrita at gmail.com> wrote:
>
>> Hi,
>>
>> I don't know if this ML is the right place to post a question about
>> NAS/iSCSI and Cluster related issue. I will write it anyway.
>>
>> info:
>> I have built a NAS server under FreeBSD. Geom RAID5 over 3 hard disks and
>> iSCSI target0 created.
>> It is up and running without problems "exporting" a SCSI driver to a Ms
>> Windows 2000 Server with Microsoft iSCSI Inititator driver.
>> The volume is NTFS.
>>
>> doubt:
>> I have read multiple "old" posts around the web that I can't connect
>> multiple initiators (w2k clients) to the same target. Is this true
>> nowadays?
>>
>> Facts:
>> I really need at least two w2k clients to the same "storage NAS". Any tip
>> about how to solve this?
>> Maybe some kind of failover technique between both w2k servers?
>>
>> I really appreciate your help. Thanks in advance
>>
>> Best Regards
>> Francisco
>>
>> PS: I sent this email to freebsd-scsi ML too. No anwsers until now.
>>
>> --
>> blog: http://sufixo.com/raw
>> http://www.linkedin.com/in/franciscocabrita
>> _______________________________________________
>> freebsd-cluster at freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-cluster
>> To unsubscribe, send any mail to "freebsd-cluster-unsubscribe at freebsd.org
>> "
>>
>
>


-- 
blog: http://sufixo.com/raw
http://www.linkedin.com/in/franciscocabrita