Notes on a Full Node

Summary

Below are a few observations on my now completely full node which has been running since early September 2019. My node just finished up 7 months, and is now in its 8th month. Allocated space is 3.5 TB on RAID 5 using an Enterprise level SAS drive array. The node is currently running version 1.0.1 in GNU/Linux Docker on Debian Stable. The last few updates have all occurred automatically and error free with watchtower. Recently, there was a short power outage. The node rebooted automatically when the power returned, and continued to operate without any issues.

All things considered, the GNU/Linux Docker setup seems to be fairly stable and reliable. From the logs, it seems that the node is no longer receiving upload requests. Perhaps this means the satellites are not including full nodes in upload requests, which is precisely what should happen.

I have no plans on extending the space allocation. So, I’ll see what life is like as an SNO with a full node.

I may start a second node running on a ZFS array… after the worldwide medical situation settles down to a quiet roar, and I have the physical and mental space to do so once the kids return to school.

Allocated space on Docker command line

-e STORAGE="3.5TB"

blobs in bytes

du -sc -hb blobs

3481903352064

available space from log

92497280

Web Dashboard Disk Space Remaining

20.10 GB

trash directory

18615787776

Actual used space in bytes

3481903352064 + 92497280 + 18615787776 = 3500611637120

Difference between allocated space and actual usage including trash directory:

3500000000000 - 3500611637120 = -611637120

About 611 MB overhead.

The node doesn’t seem to be receiving any upload requests:

$ grep upload node.log |tail

2020-04-05T10:00:57.081Z	INFO	piecestore	upload started	{"Piece ID": "XGO...RUICUXQ", "Satellite ID": "1wFTA...", "Action": "PUT", "Available Space": 102822528}
2020-04-05T10:00:57.990Z	INFO	piecestore	upload canceled	{"Piece ID": "XGO...RUICUXQ", "Satellite ID": "1wFTA...", "Action": "PUT", "error": "context canceled"...
2020-04-05T10:00:58.131Z	INFO	piecestore	upload started	{"Piece ID": "KJJ...HCK7QMA", "Satellite ID": "1wFTA...", "Action": "PUT", "Available Space": 101776768}
2020-04-05T10:01:00.166Z	INFO	piecestore	uploaded	{"Piece ID": "KJJ...HCK7QMA", "Satellite ID": "1wFTA...", "Action": "PUT"}
2020-04-05T10:01:00.563Z	INFO	piecestore	uploaded	{"Piece ID": "54T...ZZWUTKQ", "Satellite ID": "1wFTA...", "Action": "PUT"}
2020-04-05T10:01:00.738Z	INFO	piecestore	upload started	{"Piece ID": "LB4...RRPGOUQ", "Satellite ID": "1wFTA...", "Action": "PUT", "Available Space": 97137024}
2020-04-05T10:01:02.248Z	INFO	piecestore	uploaded	{"Piece ID": "EXA...RYOLECQ", "Satellite ID": "1wFTA...", "Action": "PUT"}
2020-04-05T10:01:03.315Z	INFO	piecestore	upload canceled	{"Piece ID": "LB4...RRPGOUQ", "Satellite ID": "1wFTA...", "Action": "PUT", "error": "context canceled"...
2020-04-05T10:01:54.378Z	INFO	piecestore	upload started	{"Piece ID": "HEG...OVPCTLQ", "Satellite ID": "12L9Z...", "Action": "PUT_REPAIR", "Available Space": 92497280}
2020-04-05T10:01:55.194Z	INFO	piecestore	uploaded	{"Piece ID": "HEG...OVPCTLQ", "Satellite ID": "12L9Z...", "Action": "PUT_REPAIR"}
5 Likes

Good luck on preserving it healthy and getting lots of uploads, then!

It is good idea to check on a node now and then but if you manage not touching it for long periods of time and get nice amount of coins per month that would be awesome.

Hello, thanks for your comprehensive post! I’ve heard from other people that full node also lacks a majority of egress traffic, I mean egress will decrease several times… Could you please post your evaluations on this?
In fact last days there is quite a few egress traffic, but anyway it is interesting to know - wheither you see the egress traffic drop or not.
Thanks!

2 Likes

Great overview, thanks.

A quick thought - the vast majority of data on my full node (2 of 6) is test data. If this could be replaced by actual customer data, or some insights on when and how this will be cycled or replaced by even just new test data could be a good motivation. Haven’t seen much related to this in forums?

I’ve been adding new nodes when old ones fill up, but honestly it seems very short lived and somewhat silly when new ones are just filling up with new (almost static and I requested test data) from saltlake satellite almost before a full vetting cycle is completed…

Just a couple of thoughts :slight_smile:

1 Like