[ovirt-devel] adding new paths to iscsi data storage domains

Paul Dyer pmdyermms at gmail.com
Sun Mar 15 14:38:26 UTC 2015


Hi,

1)  The storage device has 2 controllers.   Each controller has 4 nics.
All 8 nics are connected to a level 2 switch that has ethernet connections
to the hosts.   The storage device nics enumerated are 0, 1, 2, 3.
 Controller 0, nic 0 are the primary target for a group which includes
controller 1,  nic 0.   In /var/lib/iscsi/nodes/{IQN}/{nic}/default, the
difference between c0n0 and c1n0 is :
# diff 10.251.6.10?,3260,?/default
3c3
< node.tpgt = 1
---
> node.tpgt = 2
47c47
< node.conn[0].address = 10.251.6.101
---
> node.conn[0].address = 10.251.6.102

The 4 groups of nics have different subnets.  Dell docs recommend this, and
the storage devices keeps packets separated by these subnets.   For my
setup, I have this config:
             controller 0         controller 1        host-1
 host-2
nic0   10.251.6.101      10.251.6.102     10.251.6.135      10.251.6.136
nic1   10.251.7.101      10.251.7.102     10.251.7.135      10.251.7.136
nic2   10.251.8.101      10.251.8.102     10.251.8.135      10.251.8.136
nic3   10.251.9.101      10.251.9.102     10.251.9.135      10.251.9.136

On each virtualization host, I have 1 nics configured on each of the 4
subnets.

2) We are using rhev 3.5.  I have deployed this as round-robin, not bond.
According to Dell support, iscsi works best with round-robin, whereas FC
NAS works best with bond.   I follow their recommendations.   The rdac
driver is setup for this, and the prio= below separates the 8 nic paths
into 2 groups.

# multipath -ll
36f01faf000d7ddeb000002085258bce5 dm-1 DELL,MD32xxi
size=756G features='3 queue_if_no_path pg_init_retries 50' hwhandler='1
rdac' wp=rw
|-+- policy='round-robin 0' *prio=14* status=active
| |- 7:0:0:1  sdc 8:32   active ready running
| |- 8:0:0:1  sdj 8:144  active ready running
| |- 11:0:0:1 sdu 65:64  active ready running
| `- 12:0:0:1 sdv 65:80  active ready running
`-+- policy='round-robin 0' *prio=9* status=enabled
  |- 6:0:0:1  sdf 8:80   active ready running
  |- 10:0:0:1 sdk 8:160  active ready running
  |- 5:0:0:1  sdo 8:224  active ready running
  `- 9:0:0:1  sdt 65:48  active ready running

Hope this helps explain my setup.   I am not sure how to file a bug.   Is
this done on bugzilla or somewhere else?

Paul


On Sun, Mar 15, 2015 at 7:44 AM, Elad Ben Aharon <ebenahar at redhat.com>
wrote:

> Hi Paul,
>
> I would like to know the following details:
> 1) Are the hosts's NICs connected to the storage server located in the
> same network subnet as the storage server itself?
> 2) Have you tried to deploy the connection to the storage server using the
> 'iSCSI multipath' bond that available in RHEV-3.4?
>
> ------------------------------
> *From: *"Nir Soffer" <nsoffer at redhat.com>
> *To: *"Paul Dyer" <pmdyermms at gmail.com>
> *Cc: *devel at ovirt.org, "Elad Ben Aharon" <ebenahar at redhat.com>
> *Sent: *Sunday, 15 March, 2015 12:54:44 PM
>
> *Subject: *Re: [ovirt-devel] adding new paths to iscsi data storage
> domains
>
> Adding Elad who tested this feature lately to add more info.
>
> ----- Original Message -----
> > From: "Paul Dyer" <pmdyermms at gmail.com>
> > To: "Nir Soffer" <nsoffer at redhat.com>
> > Cc: devel at ovirt.org
> > Sent: Friday, March 13, 2015 6:25:05 PM
> > Subject: Re: [ovirt-devel] adding new paths to iscsi data storage domains
> >
> > Nir,
> >
> > we have added 2 more nics to each virtualization host.    In order to get
> > this working, I had to add an after_network_setup hook.   The shell
> script
> > simply does "/sbin/iscsiadm  -m node -L all", to get the extra targets
> > login after reboot.
> >
> > I looked in the engine table storage_server_connections and found that
> only
> > the iscsi targets selected during the original storage domain create were
> > present.   If ovirt-engine added rows here, then most of the work would
> > have been done.
> >
> > I say mostly, because the Dell MD3200i did not return exactly the correct
> > portal target values.   The device has 2 controllers, with 4 nics each.
> > Controller 0, ports 0,1,2,3 uses portal target 1.    Controller 1, the
> > ports uses portal target 2.   After iscsiadm discovery, the portal
> targets
> > for ports 1,2,3 were all 1.    Ports 0 had targets 1 and 2, correctly.
>  I
> > adjusted the values saved on the filesystem, and login/logoff works fine.
> >
> > Paul
> >
> >
> >
> > On Fri, Feb 6, 2015 at 11:57 AM, Paul Dyer <pmdyermms at gmail.com> wrote:
> >
> > > First of all, thank you for your time.   I must apologize that in this
> > > install, I am using RHEV 3.4.5.    I will try to reproduce this on an
> ovirt
> > > install.   I just need to create some paths to iscsi targets.
> > >
> > > 1.   This configuration has 2 physical hosts, Dell PE-R715 servers,
> with a
> > > Dell PowerVault MD3200i iSCSI data storage domain.   The EM3 nic was
> the
> > > original link.   EM4 was the new connection to storage.
> > >
> > > 2.   From the manager interface, I selected Storage tab, then the
> > > kvm5DataDomain, then edit.   From the popup, I added the IP address
> under
> > > Discovery Targets, then clicked the Discover button.    Then, clicked
> the
> > > login arrow on the new targets discovered.
> > >
> > > I have attached the engine and vdsm logs.   I was working on this at
> about
> > > 11:40am Feb 4th.
> > >
> > > When setting the host in maintenance mode, then reboot, and Activate,
> the
> > > new paths do not get a login.
> > >
> > > Thanks,
> > > Paul
> > >
> > >
> > >
> > > On Fri, Feb 6, 2015 at 5:38 AM, Nir Soffer <nsoffer at redhat.com> wrote:
> > >
> > >> ----- Original Message -----
> > >> > From: "Paul Dyer" <pmdyermms at gmail.com>
> > >> > To: devel at ovirt.org
> > >> > Sent: Friday, February 6, 2015 12:20:23 AM
> > >> > Subject: [ovirt-devel] adding new paths to iscsi data storage
> domains
> > >> >
> > >> > Hi,
> > >> >
> > >> > I have been reading the devel list for months now, and would like to
> > >> ask a
> > >> > question.
> > >> >
> > >> > In version 3.4.5, adding new paths to an existing iSCSI data storage
> > >> domain
> > >> > does not work from the manager.
> > >>
> > >> It works on ovirt 3.5 and master and it should work also in all
> previous
> > >> versions.
> > >>
> > >> Please open a bug for this:
> > >> 1. Describe the configuration you are modifying
> > >> 2. Describe the steps you take
> > >> 3. Include engine log
> > >> 4. Include vdsm log from the host trying to add new devices
> > >>
> > >> > I have been able to add the paths with
> > >> > command line "iscsiadm -m discovery -t st" and "iscsiadm -m node -L
> > >> all".
> > >> >
> > >> > Is there any plan to allow for adding new storage paths after the
> data
> > >> domain
> > >> > has been created?
> > >> >
> > >> > Thanks,
> > >> > Paul
> > >> >
> > >> >
> > >> > --
> > >> > Paul Dyer,
> > >> > Mercury Consulting Group, RHCE
> > >> > 504-302-8750
> > >> >
> > >> > _______________________________________________
> > >> > Devel mailing list
> > >> > Devel at ovirt.org
> > >> > http://lists.ovirt.org/mailman/listinfo/devel
> > >>
> > >
> > >
> > >
> > > --
> > > Paul Dyer,
> > > Mercury Consulting Group, RHCE
> > > 504-302-8750
> > >
> >
> >
> >
> > --
> > Paul Dyer,
> > Mercury Consulting Group, RHCE
> > 504-302-8750
> >
>
>


-- 
Paul Dyer,
Mercury Consulting Group, RHCE
504-302-8750
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/devel/attachments/20150315/83eddf5b/attachment-0001.html>


More information about the Devel mailing list