Why not syncing metadata exclusively in RAM?

Feature requests
#1

As all we read and experiment daily, one of the most frustrating process on Insync is the long and never-ending “syncing metadata”. My computer is a Core 2 duo with 8 GB of ram. I’ve got about 200 GB of synced data. My gd2-xxxxxxxx.db-wal has actually 427 Mbytes. Despite all this RAM, Insync uses all the time the hard disk very intensively, slowing the computer and the rest of the processes. Two questions: 1) Why Insync needs to use all the time (when syncing metadata) the hard disk even having a lot of RAM available. 2) Would be possible to change something to use the RAM for this process but not the hard drive, except when is absolutely necessary? Is it possible to load the whole database in RAM and then works in there and then save changes in hard disk when process is finished? I guess if this is possible the process of “syncing metadata” would improve the speed a lot.

This is an usual example: syncing metadata and hard disk running heavily while there is more than 60% of RAM available and CPU is barely used (no more than 12%). Why not try to use more RAM and more CPU to improve the speed of “syncing metadata”?

Insync 1.2.17.35187 for Ubuntu 14.04 - 64bit.
Data synced: 200 GB.
Files synced: 140.000.
Linux Mint 17.2 - 64bit.
Intel Core 2 Duo - 3.0 Ghz - 8 GB RAM.

#2

Hi @quasaribex, I apologize for the late response.

Can you kindly send in your logs and the link to this post to support@insynchq.com so we can investigate this?

Thanks!

#3

May I suggest that a RAM DRIVE to hold the database files accomplishes this quite well. DataRAM’s RAMDISK is what I’m using and it provides a great work-around from having insync slamming the hard drive. It also speeds things up a lot, although it will slow your machine down when insync is really busy because your hard drive is no longer a bottleneck limiting CPU access. That said, I find it works quite well.