FWIW, Virginia wasn't polled much in 2020 because it wasn't really considered competitive, but the polls that we did get were pretty accurate.
Looking at VA polling in post-Obama elections:
VA-GOV 2013: overestimated McAuliffe (D) by 3-4%
VA-SEN 2014: overestimated Warner (D) by 8%
VA President 2016: accurate to within 1%
VA-GOV 2017: underestimated Northam (D) by 5-6%
VA-SEN 2018: slightly overestimated Kaine (D) by perhaps 1-2%, very few polls because considered uncompetitive, includes stuff from several months before the election with tons of undecideds
VA President 2020: slight overestimation of Biden, perhaps 1-2%, nothing like the nationwide error
VA-SEN 2020: overestimated Warner by perhaps 4%, but the sample includes several early 5X/3X polls with tons of undecideds like in 2018 VA-SEN
This isn't like the Midwestern states where McAuliffe +4 in the polling average would mean to expect a Youngkin win, and the big errors from the early 2010's seem to have been fixed, and even repeating the 2013 error would have McAuliffe winning in a recount. However, the Dem still needs to be up by at least 2-3 in the polling average to be confident.
What's interesting is that D executive offices/R legislature split ticket voting looks very much alive. The Dem House majority would be gone at only D+1.