Forums

Home / Forums

You need to log in to create posts and topics. Login · Register

setup freeze on step 6

Pages: 1 2 3

Hello, thx, unchecking the "Local Storage Service" option actually did the installation proceed, but now I can not open the iSCSI disk settings or create a disk or manage(opens the window, but hangs when ordering to create), it will load infinitely

(Print in email)

I needed to know how many disks you have on your first node.

Before you can create iSCSI disks you need to have a pool of storage disks that can store data. These storage disks need to be separate/complete disks other than the system disks.

Assuming in your case you have 3 machines with only 1 disk each, then you need to add 1 more disk on each node ( so each will have 2 disks ) and then either:

  • Rebuild the cluster again, leaving the "Local Storage Service" selected on all three machines. This is probably the easiest way.
  • Keep the cluster that you have built then re-add the "Local Storage Service" role on all 3 nodes from the "Node List" page, then go to the "Physical Disk List" for each node, then click the "+" action to add the second disk to storage pool.

Once your cluster has storage disks ( 1 disk on each machine ), you will be able to create iSCSI disks.

My reply is based on the assumption you current problem is having only 1 disk per node. If this is not the case then you need to run the command:

ceph-disk list > result

as per my previous post so we can check if we are able to detect the storage disks.

Now everything works, the problem was that.

So does iSCSI only work if local storage is enabled?

Is there any topic or documentation to connect MPIO in linux? When I tried to do the discovery with iscsiadm, I only received "Connection refused"

Good to hear it is working now  🙂

Yes you need to have  nodes with "Local Storage Service"  selected  ( default value ) + storage disks (system disks do not count).

We do give the option to de-select this since for advanced setups it is possible to have nodes with dedicated roles: nodes that only do management, nodes that only run iSCSI services and nodes dedicated to only storage.

I can not help with the Linux iSCSI setup, all our client tests are done with Windows and WMWare/ESXi

Okay, but apparently, iSCSI is not booting, or it does not start on port 3260, even though it has the status "started". I tried a telnet on both the IP that I received in Active Path when on the cluster nodes and I could not close a connection on port 3260.

Can you ping from your client machine to the ips of the disk

If not then you need to setup your client to access the iSCSI subnets, make sure you have nics/swtches that connect your client to those subnets and that you client has IP addresses on both subnets.

Yes, i can ping the disk

If you can ping, it is most likely a configuration issue on the client side. It will help a lot if you do have a Windows machine or VMWare  ESXi to try to connect to the discs so we can confirm this.

Following  link shows you how to setup connection from linux client using open-iscsi:

https://www.suse.com/documentation/ses-3/book_storage_admin/data/ceph_iscsi_connect.html

We have not tested  linux clients ourselves but you are correct that we should. Once we do it it will be documented like the other use cases we have.

After you said that you had no way to help me with linux, I tried to do it on windows, and it was in it that I tried telnet and did not get a response, I also tried to follow the documentation, but gave error in doing the discovery, on linux I was not allowed to make the connection too.

Maybe it's the same problem for both systems

My firewall was disabled while I was testing, to see if it was not blocking it

I sent some prints in the email

Hi,

Assuming your first disk path 192.168.22.100  is served on petasan3 (note this can change) on interface ethX ( ethX is the interface mapped to the iSCSI subnet for 192.168.22.100 ) :

ssh to petasan3 and please capture the result of the following commands :

this should show if the ip path has been assigned by the OS:

ip addr | grep "192.168.22.100"

this should show that 192.168.22.100:3260 has been activated by the linux lio target:

targetcli ls

Or

targetcli ls > result

Also we want to test if it is able to receive traffic from the Windows client machine

run

tcpdump -i ethX  icmp

after it is running ping 192.168.22.100 from the Windows box, does tcpdump print any packets ?

Then test if iSCSI packets are being received:

run

tcpdump -nn -i ethX  port 3260

Then on the Winodws side retry the iSCSI discovery, does tcpdump print any packets ?

can you also send:

/opt/petasan/config/cluster_info.json

output of

dmesg

Some other questions:

-is the disk status Started or Starting ?

-the 2 paths you have for the disks are on the same subnet 192.168.22.100 and 192.168.22.101 did you specify having 2 paths on the dame subnet when you created your disk or are both iSCSI1 and iSCSI2 on 192.168.22.0 which would not correct.

Pages: 1 2 3