Considering that they are independent VC's(Either preprovisioned, VC's inconnected on non-VCP ports), they are going act to 3 independent switches and hence connected VC3 back to VC1 in a similar fashion will end up creating a loop.
For you to achieve shorted way from VC1 to VC3 and failover in case of anyone device going down, you can connect VC1 to VC3 but you must be running STP between these VC's to avoid the loops.
Hope this helps.
Thanks, Pradeep Please Mark My Solution Accepted if it Helped, Kudos are Appreciated too!!!
yesterday I connected VC1 and VC3 the same way VC1 and VC2 is connected. The good news: Nothing broke, everything kept working as expected.
But: I did some traffic tests, and our monitoring shows that now there is almost no traffic on the connection between VC1 and VC2. Even when i send data from a device in VC1 to a device in VC2 it's "routed" via VC3. Is this expected? Do I manually have to tell the Virtual Chassis the shortest paths?
The VC connections inside each VC automatically takes the shortest path. The connetions between VC1, VC2 and VC3 are handled via Rapid spanning-tree meaning that you switches will block the traffic between either VC1<->VC3, VC1<->VC2 or VC2<->VC3.
This is by design and you cannot get to a design where all links are active at the same time. STP were designed for failover redudancy, not more bandwidth/capacity.
You can change RSTP configuration on your three VC's to make the STP block happen between VC2 and VC3 if that makes more sense... but the traffic between VC2 and VC3 has to be forwarded via VC1.