You will need to read those almost all of the sub pages listed on the special upgrades. Off the top of my head, I know that you need to get to v4.8 or v4.9 before getting to v5 and then there are special steps and procedures for certain versions.
If you can take the downtime, you might be able to use asbackup and take the entire cluster offline/upgrade it and try to restore - but you’ll need to test that on your own (esp. ensuring the backup is OK before doing anything)
To be perfectly clear: I think you will have problems if you try to upgrades from v4.2 to v6.2 directly. That goes for x86 or ARM.
If the question is more geared toward x86 vs ARM, I haven’t heard or suspected any compatibility issues - as I assume the fabric and wire protocol remains the same.
I understand that upgrade in-place will be complicated ,need step by step ，one by one ; but if can i change the node one by one from 4.8 to 6.2 ,does node in different version can work together compatibly in one cluster ?
we have a cluster than running different version 184.108.40.206 and 220.127.116.11 .