1Gb RAM

For RAM, the bigger the better. After monitoring several 1 Gb RAM hotspots, I believe 1Gb RAM isn’t the bottleneck of your hotspot.

Many people are worring about the Bobcat, Helium and Rak v1 1Gb RAM hotspots after the announcement from Helium. Will the 1G RAM be the bottleneck of your hotspot? My answer is NO. The Helium core team is excellent. The Miner’s performance is amazing. It’s only using about 400Mb RAM in a normal working state.

In fact, Bobcat has shipped over 170,000 hotspots so far in which 34.3% is 1Gb RAM, and 65.7% is 2Gb RAM. The 2Gb RAM model was shipped from August 2021.

I’ve asked Bobcat Support to add several monitoring software in both the 1Gb model and 2Gb model hotspots. Let’s take a look at the memory usage.

1Gb RAM Model

The hotspot tested operated on WiFi with a stock Antenna on the roof that was about 6m high. The memory usage amounted to roughly 400mb. The max CPU load was about 2.2 (4 core CPU). CPU usage was at about 40%. No need to worry about the RAM at all.

2Gb RAM Model

Compared to 1Gb model, the 2Gb model only uses 400M memory as well. The CPU loadis only 1.2 (4 cores). If you have 4G, 8G memory, it will only use 400M too. IMO, 4G, 8G memory is kind of a waste.

Once all hotspots are changed to light gateways, RAM usage will likely decrease from 400mb to about 100mb.

When will the large RAM will be used? Loading a snapshot. The snapshot is a big file. But no matter how big your RAM is, if the snapshot is big enough, it will be out of memory if the miner program opens a large file directly and loads it all into the RAM. Of course large files won’t be loaded in the miner all at once. It will be loaded little by little. So if you have a big RAM, it can be several minutes faster to load a snapshot. Since we are not loading snapshots everyday, I don’t think loading a snapshot slower will affect our mining too much.

Issues related to Chain Halt and API

There were many issues recently. Some of these issues even caused the blockchain to stop generating new blocks. But are those issues caused by 1Gb RAM hotspots? NO.

Big Block Issue

A very big block generated at 1105523. It has about 35820 transactions in it. If you download it to local PC, the size is about 12M. The big block is obviously generated by some bugs in the network.

Normally how many transactions are there in a block? About 1000 (for now). So the big block is about 35 times bigger than a normal one. https://explorer.helium.com/blocks/1105523

It’s only 12M, so why the miner can’t process it? Keep in mind it’s not only the data itself. When data is loading in the miner, it will need to process the data and additionally communicate it to the blockchain. A lot of resources are needed to process it. The CG stopped working after this big block generated.

What caused the mega block? It is said Discovery Mode caused the mega block but it is unsure. Maybe some rate limit should be added in the miner. When running Disco Mode, there are many transferred packets. There are actually many transferred packets in some hotspots even without Discovery Mode, for example:

Region Undefined

According to the helium miner repo on Github, it looks like the current PoC V10 is still using a centralized solution to determine the region of a hotspot. It will lookup the frequency plan by the country code of your hotspot location. The API is https://api.helium.io/v1/hotspots/{pubkey}.

As you know, many people are using some dashboards to mornitor their hotspots. These dashboards will retrieve hotspot information using the same API. The API is also widely used by Explorer, App, etc. Helium is adding many new API servers everyday to handle more and more requests but it’s still not enough because the API is centralized, the DB io and network brandwidth will become the bottleneck when a large amount of requests come in.

When the API slows down, and your hotspot restarts, it will try to get the region again but it can’t because it will timeout after 5 seconds. So the miner will try again and again to get it’s region – your miner will be “region undefined” until the region is found.

Another possibility is the API side restricted the access for a period of time which also banned the miner to get data from it since there wasn’t a User-Agent defined in it as well.

PoCV11 is supposed to solve this issue.

Helium API Error

You may have seen API errors on your dashboards, disgnoser pages. Blockchain height information can’t be shown when a Height API error happens. But don’t worry about that, it’s only a display issue, it won’t affect your mining at all.

The Helium API error is also caused by too many requests to API server. The rate limit has been applied to API endpoints. Only in 20s / 60s can return a result, so don’t refresh your dashboard again and again, give it some time before another click.

As you can see, none of these issues were really caused by 1Gb RAM. So please trust Helium, they will make the network better.

Leave a Reply

Your email address will not be published. Required fields are marked *