Truenas Rysnc Help

TrueNAS General
,
1

Question for the experts:

I have finally finished setting up my primary Truenas-1 which is Baremetal, and setting up Virtualized Truenas-2 in Proxmox.

Truenas 1 has 2 onboard 1gig ports, and a 2 port 10gig X550-t2 card.

Truenas 2 has hardware pass through and all the drives and 2nd X550-t2 installed.

Both work fine and no issues by themselves.

Truenas-1 - Bare Metal is the following:
1gig port 1 - has a Static of 10.0.0.11 - primary network/Internet
1gig port 2 - setup as DHCP

10gig port 1 - has a static of 192.168.0.3, connected to Truenas-2 10gig port at 192.168.0.2

10gig port 2 - has a static of 192.168.1.3 not connected to Truenas-2 10gig port at 192.168.1.2

Truenas-2 - Virtualized:
1gig port - has a static of 10.0.0.23
2nd port is not passthrough.

Both machines see the 10gig connection to each other.

I have tried to create and run an RSYNC Module task to push/pull ALL data from one machine to the other. Every time I have tried different methods it fails regardless of which machine I tried to set the task up on.

I want to maintain a 1:1 copy of all my data.

I have not tried doind to a ssh task and need to read up on the process. As far as I am aware this should be a possible/achievable thing going peer to peer as opposed thorugh the network. The reason for this, is 120TB of data need to be migrated from one machine to another so I can rebuild the raid once I am done.

Thanks

2

Also, I am using the internal Rsync service inside Truenas, not the App. This is my next test.

3

Why are you not using replication?

4

Honestly, going over and rewatching Toms video, I am going to try that since it is Truenas to Truenas. I think the whole block level versus file level thing is throwing me off, aswell as only snapshot being transfered. I do think that the peer to peer aspect is throwing a curve into it. 10gig to 10gig bypassing the main network should work, as I have done it with Windows to Truenas.

5

  1. TrueNAS Scale only includes the Rsync client (which is used by the machine where you start the sync). The other machine needs to run the Rsync app which is the Rsync server that the Rsync client needs to connect to.

  2. @NugentS is absolutely right - you should be using replication instead. Rsync is a very chatty app - for each file the client needs to ask the server for details of the file it has, then send the file if the details have changed. Because replication is based on snapshots, it can simply stream all the changed blocks from one server to the other at maximum speed - so it will be much much quicker.

  3. Snapshots are a wonderful thing - you should be using them anyway. And replication using them is a no-brainer then.

6

This generally does not give the results you would expect. Basically, DHCP is setup as a lone / single network. If you want multiple network ports active, you must use static for all.

The reasons are numerous, but the biggest is router. Unix does support multiple routers, including host and sub-net alternate static routes. However DHCP can cause asymmetrical routing when used with static IPs and routes.

You imply that there is a 10Gbps link between the 2 servers… where?

7

I took this as the explanation of the 10Gbps network and addressing. Easy to miss

1 Like