are you cabled point to point or do you have a FC switch?
The port states should show as online before doing any config.
Does dmesg say the link is up?
Do you see anything in dmesg?
Also… p2p or using a fc switch?
I’m having an issue with it limiting my posts to this thread. I had to delete an older one to post this. Is there a work around?
P2P
Server:
Client:
targetcli - What’s missing? Still no block device visible on the initiator
-
speed/connectivity issue is probably bad/wrong/ or dirty cables
-
you dont use targetcli on the initiator/client side
try
ls -l /dev/disk/by-path/
I’m not using it on the client side. Only server side. Just showing you the responses to the things you asked.
This is awesome. I’ve never used it before.
With the FreeNAS box I have connected to the second port identified as 0x21000024ff56a8a9 on the initiator, I can see all those block devices. The two block devices from this target don’t show up at all on 0x21000024ff56a8a8 as they should. Did you see anything wrong or missing in my targetcli’s ls on the target?
sorry but I am not trying to analyze your particular setup/config in that much detail.
Ok, no problem. Thought you might see it right off. I’ll dig a bit. I followed all the steps I saw at various sites. I can’t imagine what I’m missing. Looks like the last step.
Can you share the full ls of your targetcli so I can compare? The one you have above only show the qla2xxx.
I didn’t see your backstores but comparing yours against mine, you have fileio but I did block. Should that make a difference?
I used block so that the initiator can create/change any LVMs on it.
I havent used CLVM. I dont have any suggestions about it.
But I dont know of any difference.
I am using img files , so thats why I am using fileio.
from glance, it looks like your acls arent right.
What looks off on the ACLs?
I used the names I assigned to the devices when I created the backstore. Better to use the actual device name rather than the name?
/backstores/block> create dev=/dev/sda1 name=vm_shared
/backstores/block> create dev=/dev/sda2 name=vm_backup
That’s why they show [lun0 block/vm_shared (rw)]
I haven’t found a real definitive guide for this. Yours was closest.
you put a target wwn as the client acl
The other client wwn under the other targets at least doesnt match any of the wwn targets. Which is probably why it appears there a lun there.
probably b/c this type of work isnt normally free
when you get it 100%, will you know how to setup multipath, or is your plan to use single paths to each client initiator?
1 Like
1 Like
I’m not there yet so it’s not factored into any of this right now.
To get the following, I used find /sys -name port_name | xargs head
In that last pic, naa.21000024ff67440f under qla2xxxx is the port_name of the target. Under acls, naa.21000024ff56a8a8 is the port_name on the initiator.
In your examples, naa.2100001b329a6e08 is your port_name on the target and naa.2100001b3217eb9a is the port_name on your initiator, correct?
Isn’t that how it’s supposed to be?
In this pic, does that mean the initiator is seeing the target, just not configured correctly?
possibly…
not sure why you created separate block device targets for (2)both partitions on drive sda
why not just pass the whole drive as a target?
I created one for backup space and the other one to have LVMs for VMs completely separate of each other.
What are all those fc-0x devices?
Those are from a FreeNAS box that I’m trying to cut away from.
have you rebooted?