I think that a full reindex is slower and less reliable than anyone likes to admit to.
Its needed to fix certain errors, not sure what time frame you consider "slower than like to admit".
0.12.0 has changed some stuff which might index the chain differently enough to want reindexing.
In linux at a terminal,
after [file][quit] to stop bitcoin if it had autostarted at login,
bitcoin-qt -reindex -datadir=/home/username/.bitcoin &
You dont have to pass the default datadir.
I'd expect that to run for three days straight on an average 3GHz quadcore such as an i5 with plenty (>=3GB) of RAM
Id expect <24 hours with 0.12.
If that runs for a random length of time between an hour and a day before crashes, try moving (?deleting) /.bitcoin/database/ and /.bitcoin/peers.dat and then
bitcoin-qt -reindex -datadir=/home/username/.bitcoin &
Why delete the peers.dat? Also there is no /.bitcoin/database folder, are you refering to
chainstate maybe?
Once a reindex has completed, you can start BTC as normal at login or from its icon, and it remembers the location of your /.bitcoin/
As for how to speed things up, it is indexing every BTC transaction ever. Perhaps could someone comment on pruned nodes ?
Pruned nodes dont store the entire blockchain, but only a part of it. What exactly is your question?
Anyone know if there's any way to accelerate this reindex?
Get better hardware.
I think the setting the
-dbcache=<n> where N is the size in megabytes of the cache to a larger size might make it faster.
dbcache=4096 is a speedup, but it will result in core using ~5.5 GB of memory.
I recently did a full sync in 8.5 hours[1] with 2/4 i5 cores on a mechanical disk. It significantly slows shutdown time though, so I wouldnt use that as a normal setting.
[1]
https://bitcointalk.org/index.php?topic=1357766.msg14008753#msg14008753