# Running archive node Page to summarize archive nodes setups, recommended practices and potential issues. Feel free to add info about your setup below. ## Prague setup * Hardware * Machine * Dell PowerEdge R320 * CPU * Intel Xeon E5-2430, 6 core/12 thread, @ 2.7GHz * RAM * 16 GB (4x 4GB) DDR3 * Disks * 3x Samsung 860 EVO, V-NAND SSD, 4TB for client data * Disk controller Dell PERC H710 Mini 512 MB Cache * RAID 0, stripping drives into one logical volume =~11TB usable space * 300GB HDD for system * Price * Machine with accessories - ~260€ * SSDs - ~1200€ * Software * OS: Debian 10 Buster * Client * Erigon * Updates are semi-automatized with script running as service which regularly checks for new version, downloads it, verifies signature and sends notification. User then has to manually switch to the new version * Backup not solved yet * RPC authentication not solved yet * Runs using SystemD service * Time to fully sync ~4 weeks * Issues * First we tried same machine as Berlin office (Rackable Arima NM46X 2x AMD Opteron 2214 HE 16GB RAM 9650SE-12ML), it was cheap and modest solution but we ran into issues with performance and disk controller which couldn't handle drives bigger than 2TB. Also was super loud. * H710 disk controller setup on Poweredge was quite a puzzle at first but works well. It is important part of the machine which should be paid attention before buying. ## Boulder Setup * Hardware: * Machine * System 76 Thelio * CPU * 4.6 GHz AMD 3rd Gen Ryzen 9 3900X (3.8 up to 4.6 GHz - 12 Cores - 24 Threads) * RAM * 64 GB Dual Channel DDR4 @ 3200 MHz (2x 32GB) * Disks * 2 TB NVMe Seq Read: 3500MB/s, Seq Write: 3300MB/s (for system) * 12 TB SSD Seq Read: 560 MB/s, Write: 530 MB/s (3 x 4 TB) (for chain data) * Price * $4280 USD * Software * OS: Pop!_OS 20.10 (64-bit) (Ubuntu based) * Client * Geth * config: https://gist.github.com/pipermerriam/adee4c943ff9dee64413ab1d8a791858 * no backup configuration * runs using SystemD service * Time to fully sync ~30 days ## Chase Wright and DefiDude setup Archivenode.io runs multiple nodes with different setups. * Hardware: * Machine * 2U Rack Mount server with SATA SSDs * Intel NUC with an external Thunderbolt storage array * This setup had problems, has to change * Software * Windows Server 2019 * Clients * Geth * Turbo-geth * Strongly prefers, easiest and cheapest solution * Nethermind * NGINX, HAProxy, and Dshackle for hosting RPC * Another node servers as backup * Recommendations * ECC RAM * CPU - IPC matters, fastest possible * Had issues with NVMe, SATA SSDs are easier to setup * Not QLC, quality matters * RAID * Uses RAID10 (more expensive option) and RAID0 * zfs might be interesting solution * Crashes * Sudden crash and client corrupting its database is semi common issue * Disk redundancy is pointless, the nodes are far more likely to corrupt themselves from a crash than a disk failure is to occur <details> <summary> Title </summary> More info </details> | Upgrade | URGE | Track | Topic | Cross-references | |:-------:|:-----------:|:---------:|:---------------------------------:|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------:| | ePBS | the Scourge | MEV track | Endgame block production pipeline | intersection with: [ET](https://ethresear.ch/t/execution-tickets/17944), [PEPC](https://efdn.notion.site/PEPC-FAQ-0787ba2f77e14efba771ff2d903d67e4), [IL](https://eips.ethereum.org/EIPS/eip-7547) |