In my Vsphere client, I only entered the virtual ip of my Open-E cluster in dynamic discovery for iscsi targers. However, in the static discovery, it shows the ip of all other ip's on the Active Node Open-e Server. When I go to manage paths, it shows the virtual IP and a physical ip on the physical ip on the management subnet. (on one target, it is only getting the path from the physical ip, and not the virtual ip). Aside from the one Lun that does not have the virtual ip, is this normal?
In the VMware iSCSI initiator try with only using the Static, so remove the dynamic discovery and static then apply and close then open again and enter the Static IP of the Virtual IP Address.
Upon further review, it looks like this is by design and probably is not hurting anything. I think the only time this caused a problem is when I started the iscsi service on both of my open-e nodes.
In failover configuration, only one node is active at a time, so therefore the initiators should only connect to one node at a time, using the Virtual IP.
I see now that I accidentally setup multi-pathing, which is good. I don't think there are any problems.
MPIO to one node is a good thing.
Initiators connecting to both the active node, AND the inactive node, is NOT a good thing. This will result in data differences, breaking consistancy for failover.
I was worried about that, but each vhost has a different Active I/O path. One has the active I/O on the node's physical ip and the other host is to the virtual ip.
This is OK.Quote:
Originally Posted by ccrichard
In future we will have ability to designate which IP is broadcasting the target, so MPIO can be controlled easier.
Hi, when there will be a solution??
We have multipath with 4 Nics and 4 virtual IPs.
ISCSI discovery shows 8 Targets. The physical and the virtual IPs.
How can I disable ISCSI on the physical IPs??
Thank you for the help.
Regards,
Erik
Based on the limited information that you have provided I would assume you are using VMware and I would remove all Dynamic entries of the IP addresses and use only Static and if you have an HA setup w/ VMware make sure the other ESX servers are cleaned out as well before entering in the Static IP's.
Oh sorry,
I don't use esx. Ubuntu server is the client (initiator).
I use iscsiadm -m discovery -t sendtargets -p "ip".
I was hoping that you can set that only virtual IPs are found.
Perhaps I should open a new task.
Regards,
Erik
See if this helps.
http://blog.open-e.com/multipath-whi...uld-we-choose/
Yes, I use this blog post for configuration of multipath.
I think my problem is a step earlier.
lsscsi
[77:0:0:0] disk SCST_BIO mJYVlpW7lNGslAs5 200 /dev/sdg
[78:0:0:0] disk SCST_BIO mJYVlpW7lNGslAs5 200 /dev/sdl
[79:0:0:0] disk SCST_BIO mJYVlpW7lNGslAs5 200 /dev/sdk
[80:0:0:0] disk SCST_BIO mJYVlpW7lNGslAs5 200 /dev/sdj
[81:0:0:0] disk SCST_BIO mJYVlpW7lNGslAs5 200 /dev/sdi
[82:0:0:0] disk SCST_BIO mJYVlpW7lNGslAs5 200 /dev/sdm
But I only have 3 Nic's for ISCSI!!
iscsiadm -m discovery -t sendtargets -p 192.168.110.10
192.168.103.10:3260,1 iqn.2011-05:data0.target0
192.168.103.10:3260,1 iqn.2011-05:data0.target0
192.168.103.10:3260,1 iqn.2011-05:data0.target0
192.168.103.10:3260,1 iqn.2011-05:data0.target0
192.168.130.10:3260,1 iqn.2011-05:data0.target0
192.168.130.10:3260,1 iqn.2011-05:data0.target0
192.168.130.10:3260,1 iqn.2011-05:data0.target0
192.168.130.10:3260,1 iqn.2011-05:data0.target0
192.168.110.10:3260,1 iqn.2011-05:data0.target0
192.168.110.10:3260,1 iqn.2011-05:data0.target0
192.168.110.10:3260,1 iqn.2011-05:data0.target0
192.168.110.10:3260,1 iqn.2011-05:data0.target0
192.168.120.10:3260,1 iqn.2011-05:data0.target0
192.168.120.10:3260,1 iqn.2011-05:data0.target0
192.168.120.10:3260,1 iqn.2011-05:data0.target0
192.168.120.10:3260,1 iqn.2011-05:data0.target0
192.168.140.10:3260,1 iqn.2011-05:data0.target0
192.168.140.10:3260,1 iqn.2011-05:data0.target0
192.168.140.10:3260,1 iqn.2011-05:data0.target0
192.168.140.10:3260,1 iqn.2011-05:data0.target0
192.168.102.10:3260,1 iqn.2011-05:data0.target0
192.168.102.10:3260,1 iqn.2011-05:data0.target0
192.168.102.10:3260,1 iqn.2011-05:data0.target0
192.168.102.10:3260,1 iqn.2011-05:data0.target0
192.168.0.50:3260,1 iqn.2011-05:data0.target0
192.168.0.50:3260,1 iqn.2011-05:data0.target0
192.168.0.50:3260,1 iqn.2011-05:data0.target0
192.168.0.50:3260,1 iqn.2011-05:data0.target0
192.168.100.10:3260,1 iqn.2011-05:data0.target0
192.168.100.10:3260,1 iqn.2011-05:data0.target0
192.168.100.10:3260,1 iqn.2011-05:data0.target0
192.168.100.10:3260,1 iqn.2011-05:data0.target0
192.168.105.10:3260,1 iqn.2011-05:data0.target0
192.168.105.10:3260,1 iqn.2011-05:data0.target0
192.168.105.10:3260,1 iqn.2011-05:data0.target0
192.168.105.10:3260,1 iqn.2011-05:data0.target0
192.168.101.10:3260,1 iqn.2011-05:data0.target0
192.168.101.10:3260,1 iqn.2011-05:data0.target0
192.168.101.10:3260,1 iqn.2011-05:data0.target0
192.168.101.10:3260,1 iqn.2011-05:data0.target0
192.168.0.50 is Management IP
192.168.105.10 Replication IP
192.168.100.10 - 192.168.103.10 physical IP's of "data" nic's
192.168.110.10 - 192.168.150.10 virtual IP's of "data" nic's for active/passive failover
Thats my problem.
I was hoping that you can set that only virtual IPs are found.
Regards,
Erik
Try the Allow and Deny for the Target.
CONFIGURATION -> iSCSI target manager -> Targets -> target[nr] -> Function: Target IP access.
We have somehow the same Problem (DSSV6 up90 b5845):
Our ESX Hosts receive all ip-adresses for the targets including
all physical interfaces
(three DSS pairs, each DSS 6 physical interfaces) from our DSSs,
although we restricted access to the targets on DSS to the virtual ip subnets.
But our ESX-hosts do not have ip adresses configured for storage
in the physical address range of the DSSs interfaces, so there's no way for
the ESX hosts to connect to a physical interface of the DSSs.
So far, so good,
but if we have to rescan our ESXs iSCSI vmhba, it takes AGES to complete the rescan.
We suppose, that this is because the ESX tries to reach the physical interfaces
of the DSSs somehow and after a time runs finally in a timeout.
So the question is:
Is there any way to limit the "sendtargets" of DSS to the virtual interfaces as mentioned by Gr-R ?
We know, that we could configure all targets on all ESX hosts static to avoid this,Quote:
Originally Posted by Gr-R
but that's a lot of work for 6 ESX hosts with 15 DSS targets with each 3 paths....
Regards,
Ralph