[sheepdog-users] sheepdog tgt multipath

[Balázs Kónya]

Hi!

I tested your advice!

The whole concept still not working, but.

Format the cluster with -l option and tgtd helped to setup multipathing
when only one XenServer node connected to the storage.
I've not tested with VM, but checking XenServer path status, seems to
working as expected.
If i restart one sheep node with the tgtd...path goes down...after
recovery...path recovers too.

After i created a XenServer pool with two xen node, and connect the
storage...everything seems to be ok, but when i enable the XenServer High
Avaliability, only one iSCSI path remain active...

I'm sure, this is not the Sheepdog Cluster or tgt problem.

Thank You for the advice again!

2015-04-21 14:24 GMT+02:00 Balázs Kónya <brokerka at gmail.com>:

> Hi!
>
> Thank You for the fast response!
>
> Yes, i will try it, and report the results!
>
> On XenServer (Citrix 6.5) i found a good pdf to setup multipath support,
> so that will (maybe) be ok.
>
> 2015-04-21 14:13 GMT+02:00 Hitoshi Mitake <mitake.hitoshi at gmail.com>:
>
>> On Tue, Apr 21, 2015 at 8:00 PM, Balázs Kónya <brokerka at gmail.com> wrote:
>> > Hello!
>> >
>> > First of all, sorry for my english! I'm wokring on it :)
>> >
>> > I want to set up sheepdog with tgt and multipath support.
>> > I'm a newbie.
>> >
>> > On Debian Wheezy, build 3 nodes of sheepdog, the latest from git, and
>> tgt
>> > too.
>> >
>> > Everything working great, but i dont find any usabe "how-to" to setting
>> up
>> > the multipath.
>> > I want to connect the sheeep cluster to XenServer with iSCSI
>> >
>> > XenServer discover the single iSCSI path and LUN, and use it correctly.
>> > But if one of my sheepdog node going to sleep with the fishies...after
>> > recovery....
>> > The corrupted node going online, recovery is success, and iSCSI resource
>> > starting on an other node...sheepdog, and iSCSI HA seeming correct
>> (thanks
>> > to pacemaker, corosnyc)...but..
>> >
>> > Every time my VM's virtual disk become corrupted, but not the VDI inside
>> > sheepdog cluster..
>> > dog vdi check finishing successfully....
>> >
>> > The mailfuncion become inside the virtual machine...
>> >
>> > i found this article:
>> > https://www.mail-archive.com/sheepdog@lists.wpkg.org/msg13589.html
>> >
>> > This article descibes the same problem that i have!
>> >
>> > Based on this article, my sheepdog, and tgt is the latest...so multipath
>> > supporting is possibly ok..
>> >
>> > But how can i setup it, to work properly?
>>
>> Hi Balázs,
>>
>> At first, you need to format your cluster with "dog cluster -l" on the
>> latest master. It is not documented, sorry!
>>
>> And the iSCSI feature of sheepdog doesn't depend on pacemaker (and the
>> combination might not work).
>> You just need to prepare two tgtd on two hosts, login with iSCSI
>> initiator with it, and use them in fail over policy.
>>
>> In a case of multipathd of linux, you need to configure it with a
>> configuration file like below:
>> multipaths {
>>         multipath {
>>                 wwid                    $SCSI_ID
>>                 alias                   sheepdog-iscsi
>>                 path_grouping_policy    failover
>>                 path_checker            readsector0
>>                 path_selector           "round-robin 0"
>>                 failback                manual
>>                 rr_weight               priorities
>>                 no_path_retry           5
>>         }
>> }
>>
>> # you need to specify scsi id to the $SCSI_ID
>>
>> I don't know how to setup iSCSI multipath on XenServer, so I cannot
>> say anything about it :(
>>
>> Could you try the above procedure?
>>
>> Thanks,
>> Hitoshi
>>
>> >
>> > --
>> > sheepdog-users mailing lists
>> > sheepdog-users at lists.wpkg.org
>> > https://lists.wpkg.org/mailman/listinfo/sheepdog-users
>> >
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.wpkg.org/pipermail/sheepdog-users/attachments/20150421/3362e600/attachment-0005.html>