Set it to 500 manually, did not help sadly.
I'm running aarch64 on the pi3 if that is interesting in some extent.
Topic: [Raspberry Pi][Full Node][NFS] bitcoind keeps eating bandwidth after its up (Read 380 times)
May 27, 2017, 02:29:50 PM
May 27, 2017, 01:51:56 PM
Desktop has 16GB.
Raspberry pi3 has the 1GB standard and i have gpu_mem=16 in /boot/config.txt to allow maximum RAM for the node.
I also have an extra 1GB Swap file derived from the Pi's SD-card.
But the memory is far from full.
An old launch option was -dbcache=50 but I still had this issue, i'll try to play around with this value a bit more and see if it affects it at all. Thnx for the tip.
EDIT: Tried setting -dbcache=4 (which is minimum) but no difference. Still keeps downloading something at ~92 Mbit/s
Raspberry pi3 has the 1GB standard and i have gpu_mem=16 in /boot/config.txt to allow maximum RAM for the node.
I also have an extra 1GB Swap file derived from the Pi's SD-card.
But the memory is far from full.
An old launch option was -dbcache=50 but I still had this issue, i'll try to play around with this value a bit more and see if it affects it at all. Thnx for the tip.
EDIT: Tried setting -dbcache=4 (which is minimum) but no difference. Still keeps downloading something at ~92 Mbit/s
Try maximizing the database cache. The default setting on the builds for the pi could be much lower than the default for x86 builds for some reason. You could also try iotop (available in the repositories using sudo apt-get) for more details.
May 27, 2017, 01:38:45 PM
Desktop has 16GB.
Raspberry pi3 has the 1GB standard and i have gpu_mem=16 in /boot/config.txt to allow maximum RAM for the node.
I also have an extra 1GB Swap file derived from the Pi's SD-card.
But the memory is far from full.
An old launch option was -dbcache=50 but I still had this issue, i'll try to play around with this value a bit more and see if it affects it at all. Thnx for the tip.
EDIT: Tried setting -dbcache=4 (which is minimum) but no difference. Still keeps downloading something at ~92 Mbit/s
Raspberry pi3 has the 1GB standard and i have gpu_mem=16 in /boot/config.txt to allow maximum RAM for the node.
I also have an extra 1GB Swap file derived from the Pi's SD-card.
But the memory is far from full.
An old launch option was -dbcache=50 but I still had this issue, i'll try to play around with this value a bit more and see if it affects it at all. Thnx for the tip.
EDIT: Tried setting -dbcache=4 (which is minimum) but no difference. Still keeps downloading something at ~92 Mbit/s
May 27, 2017, 01:34:05 PM
How much RAM does your desktop have? The Pi could be relying on the share because it doesn't have sufficient RAM to keep a good cache of the data it needs from the block store.
May 27, 2017, 12:13:06 PM
Hello,
I have been fault-tracing something fishy that is going on when i run a node on my raspberry pi but I don't seem to get any further on my own.
The problem:
After the blocks have been indexed/verified etc. the bitcoind process keeps downloading something constantly at around ~92Mbit+/s. I have verified this both using vnstat, bwm-ng locally and on the NAS itself on that side using similar tools so I am 100% that the node is requesting some data constantly from the datadir while the service is up and running.
My setup:
Raspberry Pi 3 running 0.14.1 daemon mounts a NFS share from my NAS to store the data.
I've tested adapted launch options for the rpi such as -dbcache etc. but during the more detailed fault tracing I was just using:
$ bitcoind -datadir=/mnt/bitcoin -printtoconsole
I downloaded bitcoind to my desktop to try to recreate the problem but it works flawlessly there.
It had the exact same circumstances, same NFS share with the same data on it.
It verifies the blocks (a lot faster though since it's a relatively fast computer compared to the pi3) and then just sits there pulling almost no data from the NFS share at all (like it should). Verfied using vnstat, bwm-ng locally and on the NAS itself as before.
It doesn't seem to be an NFS issue, the mounts are mounted exactly the same on both machines during seperate testing.
Could the PI3 not be processing something fast enough that it needs to keep reading something from the datadir?
What could possibly be wrong here?
I really want to run this full node (without constantly draining my NASes bandwidth)
EDIT: The console/outputfile(s) mentions nothing suspicious from what I can tell
I have been fault-tracing something fishy that is going on when i run a node on my raspberry pi but I don't seem to get any further on my own.
The problem:
After the blocks have been indexed/verified etc. the bitcoind process keeps downloading something constantly at around ~92Mbit+/s. I have verified this both using vnstat, bwm-ng locally and on the NAS itself on that side using similar tools so I am 100% that the node is requesting some data constantly from the datadir while the service is up and running.
My setup:
Raspberry Pi 3 running 0.14.1 daemon mounts a NFS share from my NAS to store the data.
I've tested adapted launch options for the rpi such as -dbcache etc. but during the more detailed fault tracing I was just using:
$ bitcoind -datadir=/mnt/bitcoin -printtoconsole
I downloaded bitcoind to my desktop to try to recreate the problem but it works flawlessly there.
It had the exact same circumstances, same NFS share with the same data on it.
It verifies the blocks (a lot faster though since it's a relatively fast computer compared to the pi3) and then just sits there pulling almost no data from the NFS share at all (like it should). Verfied using vnstat, bwm-ng locally and on the NAS itself as before.
It doesn't seem to be an NFS issue, the mounts are mounted exactly the same on both machines during seperate testing.
Could the PI3 not be processing something fast enough that it needs to keep reading something from the datadir?
What could possibly be wrong here?
I really want to run this full node (without constantly draining my NASes bandwidth)
EDIT: The console/outputfile(s) mentions nothing suspicious from what I can tell
Jump to: