Earnings from mining directly depend on the duration of the devices that perform calculations to maintain the operation of a particular blockchain.
In this regard, it is necessary to ensure the smooth operation of mining rigs with a stable hash rate. The operation of computer equipment in 24/7 mode should be automated as much as possible and provide a high level of security.
This article discusses the problem of ensuring long uptime of weeks / months when mining on rigs with video cards.
What should be done to ensure the smooth operation of the computer when mining?
To ensure the stable long-term operation of the rigs, the following requirements must be met:
- Eliminate (minimize) the possibility of freezes and crashes;
- Automate the procedure for rebooting / restarting and restarting the computer in the event of a freeze, failure, after a power / Internet loss;
In addition, it is important to provide remote monitoring and network access to the rigs.
Let's take a closer look at how to ensure more stable operation of rigs around the clock.
Eliminate (minimize) freezes and crashes of mining rigs
To ensure stable operation of the rig you need:
- use reliable software that contains a minimum of errors (operating system, drivers, miner). To minimize the occurrence of possible failures, you need to configure the system: disable unnecessary services in the system, remove unused programs, cancel unnecessary automatic maintenance / update tasks. In case of power failures and frequent reboots of the rig, drivers may crash, which will lead to freezes and errors. In an OS like Linux, such problems rarely appear, so the rigs should be transferred to Linux-like systems (if possible);
- use a high-quality power supply unit capable of delivering the required power and voltage for a long time. The power supply must work stably during voltage surges in the network, filter impulse noise, have protection against short circuits, overloads and overheating;
- provide high-quality contacts, especially along the signal path between video cards, risers and the motherboard. It is necessary to take into account the fact that a rig with six video cards connected has tens of times more unreliable contacts through the connectors than a regular computer with one video card inserted into the PCI-E slot of the motherboard. Poor contact even in one place can lead to freezing of the entire system. Poor contact on the power line (which often happens when using adapters) can lead not only to freezing, but also to fire (because of this danger, you should not use a riser powered by a PCI-E connector). When using adapters, you need to regularly check them for heating, changes in the color of the insulation on the conductors. If the wires get too hot, then you need to replace them with more powerful ones (thicker);
- do not use too much overclocking of video cards. Chasing an extra penny can turn into a loss of dollars due to downtime caused by the termination of the work of rigs caused by failures due to overclocking / excessive downvolting. After assembling the rig, you need to find the frequencies and voltage at which it works relatively stably, but for practical work, set these values at the level of 80-90% of those found (slightly reduce the overclocking frequency). This will ensure more reliable operation of the rig and save it from unnecessary freezes / crashes. In addition, it should be borne in mind that over time, memory chips show degradation due to the effect of electromigration. Because of this, for stability, it is periodically necessary to slightly reduce the overclocking from memory.
Riga will be of great benefit to ensure round-the-clock operation watchdog timers... They are inexpensive and easy to set up and should be used on all rigs.
In addition to hardware watchdog timers, it is worth adding a software function to restart the miner or restart the computer in a batch file.
For this, the body shirt is modernized as follows:
@echo off: loop commands to run the miner (main batch file) timeout / t 1> nul taskkill / f / im miner_name> nul goto loop
Such an upgrade allows for a cyclic restart of the miner after software failures in its operation.
If the rig is unstable even without overclocking and downvolting, a good power supply is used, normal contacts are provided, then the cause of freezes and failures should be sought in the incompatibility / malfunction of the hardware used.
Perhaps the computer's RAM is malfunctioning (you need to run the memory test), the BIOS is set too high exchange rate on the PCI-E line (you need to use GEN1 or GEN2), the paging file size is insufficient, or the hard disk on which it stands is malfunctioning.
In addition to failures, there are external factors that lead to rig downtime. These include power outages and internet connection failures.
For mining, it is better to use a wired connection of the rigs to the Internet. This reduces the percentage of lost shares (stale shares) by 0,5-1%, improves ping to the pool and makes the work of miners more stable.
With an unstable 220V network, you can connect rigs via UPS, as well as use hardware watchdogs. Also in the BIOS, you need to enable the option to automatically turn on after a power failure.