Can I reduce the bandwidth demand of the sync process somehow?
Yes:
1. Download the blockchain.raw bootstrap and import it using blockchain_import
or
2. Pick one peer and use --add-exclusive-node
peer --p2p-bind-ip 127.0.0.1 for initial sync, then restart normally.
There is a certain amount of redundant downloading from multiple peers (specific depend on network conditions, your node performance, etc.). Either of the above will eliminate it.
So it takes 40 minutes to sync 6,500 blocks? Is that normal?
That is absolutely not normal, especially with an SSD. Are you sure you don't have something else using up CPU/RAM. I can sync almost the whole chain in 40 minutes on a mobile i7.
Thanks for your tips.
RAM: 10 GB free, CPU: 4 cpus at 10% load on avg.
I just started using
bitmonerod --out-peers 1 --add-exclusive-node 219.159.104.219:59704 --limit-rate-down 6000 --db-sync-mode fast:sync:10000
It still syncs from different peers (using "--p2p-bind-ip 127.0.0.1" resulted in no network activity, I had hope for that one helping to limit connections to 1 peer), but now the log looks different. I think using snychronous db-sync-mode did the trick:
2016-Mar-05 10:45:06.376823 [P2P8][62.210.245.87:33885 INC]Synced 93601/983931
2016-Mar-05 10:45:10.431624 [P2P6][62.210.245.87:33885 INC]Synced 93801/983931
2016-Mar-05 10:45:15.197708 [P2P1][62.210.245.87:33885 INC]Synced 94001/983930
2016-Mar-05 10:45:22.764763 [P2P7][62.210.245.87:33885 INC]Synced 94201/983930
2016-Mar-05 10:45:29.579743 [P2P6][62.210.245.87:33885 INC]Synced 94401/983930
2016-Mar-05 10:45:35.793358 [P2P9][62.210.245.87:33885 INC]Synced 94601/983931
2016-Mar-05 10:45:43.844316 [P2P9][62.210.245.87:33885 INC]Synced 94801/983931
2016-Mar-05 10:45:52.093073 [P2P2][62.210.245.87:33885 INC]Synced 95001/983930
2016-Mar-05 10:45:53.243136 [P2P3][73.8.137.244:55437 INC]Sync data returned unknown top block: 95001 -> 981505 [886504 blocks (615 days) behind]
SYNCHRONIZATION started
2016-Mar-05 10:45:58.458835 [P2P5][173.255.220.42:4587 INC]Sync data returned unknown top block: 95001 -> 449346 [354345 blocks (246 days) behind]
SYNCHRONIZATION started
2016-Mar-05 10:45:58.520588 [P2P8][123.3.223.110:37287 INC]Sync data returned unknown top block: 95001 -> 983933 [888932 blocks (617 days) behind]
SYNCHRONIZATION started
2016-Mar-05 10:46:03.703374 [P2P8][62.210.245.87:33885 INC]Synced 95201/983933
2016-Mar-05 10:46:17.463863 [P2P3][62.210.245.87:33885 INC]Synced 95401/983933
2016-Mar-05 10:46:18.889231 [P2P7][41.164.163.114:52465 INC]Sync data returned unknown top block: 95401 -> 983933 [888532 blocks (617 days) behind]
SYNCHRONIZATION started
2016-Mar-05 10:46:30.264306 [P2P4][62.210.245.87:33885 INC]Synced 95601/983933
2016-Mar-05 10:46:47.128749 [P2P8][62.210.245.87:33885 INC]Synced 95801/983933
Looks like the network was my bottleneck and how this is solved (no more parallel download)?
I can see 3 bitmonerod processes in iotop netting about 8 MB/s disk writes (averaged over 20s interval).
Seems to be looking much better now.
Not sure, though. Any info/hints are still appreciated.
(I will benchmark sync speed now and if it's still slow, I might resort to downloading a blockchain.raw)