Over the last few weeks I have been working on the networking setup that is much more reliable (with added security) for a homelab hosting Hive nodes.
It feels like it's a bit late to replace the ISP router with a DIY based, but here we are.
For $119, I have opted for a fanless G30b Mini PC that you can pick up on AliExpress. As for memory and storage, ADATA's 4GB DDR4-2666 RAM and 128GB SX6000 Lite for $12 and $14 respectively are more than enough to run OPNSense on, which are also the cheapest I can find that are compatible.
Here are the final specs of the router:
Intel Celeron N5100 (4C4T, 1.1 GHz base, 2.8 GHz boost)
4x Intel i225-V 2.5G NIC
4GB DDR4-2666 RAM
128GB ADATA SX6000 Lite NVMe SSD
As for the network switch, I went for the TP-Link TL-SG108E 8-port Gigabit switch for $19. This is a smart switch which allows tagging VLANs whenever appropriate to segregate the network for improved security.
Not related to the setup is the wireless access point in order to reach feature parity with off-the-shelf routers. The D-Link DRA-2060 costs me $21 so that other devices can connect the network over Wi-Fi.
The whole setup costs a grand total of $185. While not cheap, no other router at this price (or even more expensive ones) beats the OPNSense router in all aspects, including hardware and software. None of them are even close to the Celeron CPU, 4GB DDR4 RAM and let alone storage. Most routers stop receiving firmware updates after a few years upon release while this will continuously receive OPNSense updates as long as the project is active. At this point I have no idea how I got away with running nodes on standard ISP routers for several years till now.
As for resource utilization, the router uses less than 10% CPU most of the time and half of its available RAM. The extra few watts in power consumption is negligible relative to the advantages over the setup.
The techcoderx.com API node now uses Deathwing's Drone JSON-RPC reverse proxy to benefit from improved performance of API calls.
Since then, CPU usage of the VPS running the reverse proxy have fallen by half on average even with relatively low amount of traffic. Redirecting traffic of specific HAF apps (i.e. Hivemind) are noticeably better when the said HAF app is down on the node for replay.
Current rank: 95th (active rank 88th)
Votes: 4,396 MVests
Voter count: 307
Producer rewards (7 days): 27.947 HP
Producer rewards (30 days): 142.278 HP
Missed blocks (all-time): 30
block_log file size (compressed): 415 GB
block_log.artifacts file size: 1.7 GB
shared_memory.bin file size: 21 GB
All HAF apps (including Hivemind) belong to individual schemas in a single PostgreSQL database. This section shows the sizes of each schema in the database using the following query:
SELECT schemaname,
pg_size_pretty(SUM(pg_total_relation_size(relid))) AS total_size,
pg_size_pretty(SUM(pg_table_size(relid))) AS table_size,
pg_size_pretty(SUM(pg_indexes_size(relid))) AS indexes_size
FROM pg_catalog.pg_statio_user_tables
GROUP BY schemaname;
schemaname | total_size | table_size | indexes_size
--------------+------------+------------+--------------
hivemind_app | 684 GB | 360 GB | 324 GB
hive | 3002 GB | 1789 GB | 1213 GB
hafah_python | 16 kB | 16 kB | 0 bytes
(3 rows)
RAM usage: 47 GB
Compressed disk usage: 1.7 TB
Compression ratio: 2.22x