Bitcoin Forum

Bitcoin => Bitcoin Technical Support => Topic started by: vapourminer on February 15, 2017, 10:55:12 PM



Title: reindexing blocks on disc
Post by: vapourminer on February 15, 2017, 10:55:12 PM
using core 0.12.1

my computer locked up on a blockchain sync (it was behind a couple days). fired it back up and got a database error. so now its reindexing the blocks on the disc.

this seems to take a LONG time, its already been a couple hours and its maybe 1/5 of the way through.

i have a 3570k @ 4.3, 12 gb ram and the blockchain is on its own dedicated 256 gb ssd.

so the question is: is this speed normal? I seem to remember reindexing going a lot faster in earlier versions of core.


Title: Re: reindexing blocks on disc
Post by: achow101 on February 15, 2017, 10:57:35 PM
Previously there was a lot less data to crunch through but the blockchain has grown significantly in the past couple of years.

You can use the -dbcache=<n> startup option to increase the speed. IIRC the default is 300 mb, but you can set it to much higher so the reindex will go much quicker if you allocate it more RAM with that option.


Title: Re: reindexing blocks on disc
Post by: vapourminer on February 15, 2017, 11:08:03 PM

You can use the -dbcache=<n> startup option to increase the speed. IIRC the default is 300 mb, but you can set it to much higher so the reindex will go much quicker if you allocate it more RAM with that option.

gave it dbcache a value of 48000, disc queue value went from 1 to 5 on the blockchain ssd and read speed went way up. so it definitely seems to be doing something.

thanks!