If you filled say a 8TB node on a regular CMR drive… then shut it down and moved it all-at-once to a 8TB SMR HDD… would you expect it to still perform OK (since it’s not still growing at 500-1000GB/month)?
Or would the small churn of trash->new-ingress still cause problems?
I am running a node on a 6TB Seagate SMR drive for years now. Filewalker takes a lot of time, but so far it runs fine.
I think there are differences in how good different models perform. Some might work, others not so well…
I feel it will work fine: amount of churn would correlated with current deletion rate, and people complain about how long it takes to reduce the node size – specifically because of the slow churn. So I’d say, it’s well within the capability of the smr disk to absorb.
However copying your data to that disk will take a while… I’d say, at least a week, if cloning the disk. Which means you can’t clone it – you’ll have to keep the node online, and take the multiple rsync approach. How many months will that take I won’t try to guess.
As both are 8TB, the easiest would be to use a copy Station which would take only 2 or 3 days.
And for SMR I would ever move DBs on SSD and then its fine.
SMR usually does quite okay with sequential writes, therefore dd / cloning would be the more attractive option. Most of them have a sequential speed of 100-150MB/s, boiling down to less than 24 hours offline for a 8TB drive. That’s not so bad…
Rsync / cp / robocopy on the other hand…
I would run a full optimisation on the cmr, with ultradefrag or so and then clone it.
Umm take bets what happens first, problems or drive dying? Id say 50/50