In order to facilitate the inspection, it is recommended that the hostname naming format is:
Miner-3
Daemon-4
WorkerP1P2-5
WorkerC2-6
Note: The following number is the last segment of the LAN IP. If the network segments are different, such as 192.168.1.7 and 192.168.2.7 exist, then Named Miner-1-7, Miner-2-7, etc.
New machine inspection:
disable swap of all machines (Miner and Worker);
disable automatic update of Ubuntu;
disable automatic update of graphics card driver;
edit /etc/apt/sources.list file, modify the update source to Alibaba Cloud, Tencent Cloud or 163 mirror image.
Clock verification:
ln -sf /usr/share/zoneinfo/Asia/Shanghai /etc/localtime
ntpdate ntp.aliyun.com
NVME SSD batch group Raid0, mount, set the permissions of the mount directory; update
the mount path;
Start drive letter;
update the number of --raid-devices;
update the number of partitions;
update the user name and group;
copy the proof parameters to Deamon, Miner and C2-worker;
recharge the Worker wallet;
environment variable:
export LOTUS_MINER_PATH=/home/xtang/nvme_disk /lotusminer/
export LOTUS_PATH=/home/xtang/nvme_disk/lotus/
export BELLMAN_CPU_UTILIZATION=0.875
export FIL_PROOFS_MAXIMIZE_CACHING=1
export FIL_PROOFS_USE_GPU_COLUMN_BUILDER=1 # 使用GPU.
export FIL_PROOFS_USE_GPU_TREE_BUILDER=1 使用GPU.
export FIL_PROOFS_PARAMETER_CACHE=/home/xtang/nvme_disk/proofs_parameter # > 100GiB!
export FIL_PROOFS_PARENT_CACHE=/home/xtang/nvme_disk/proofs_parent # > 50GiB!
export TMPDIR=/home/xtang/nvme_disk/tmp # 封装临时文件.
Daily operation status monitoring of hardware equipment
Wallet balance check, timely reminder to recharge lotus-miner actor control list
lotus daemon block synchronization check lotus sync wait
dropped lotus-worker check
lotus-miner storage list
lotus-miner sealing workers | grep hostname | sort -k4: Whether there is a disable status
Check and clear message blockage
lotus mpool pending -local | grep Message | wc -l
scheduler work problem troubleshooting
lotus-miner sealing jobs and lotus-miner sealing workers | grep hostname
space-time proof daily check
Miner machine Proof file (whether the file exists in the proof file directory)
Whether the account of the time-space proof is blocked by messages
Use lotus sync wait on the node machine to check whether it is synchronized to the latest height
Check the graphics card driver, nvidia-smi to see if the driver is normal
Check the Miner process (lotus- Miner sealing job to see if the process exists)
Abnormal state sector processing lotus-miner info
PreCommitFailed: lotus-miner sectors remove --really-do-it <sectorId>
SealPreCommit1Failed: lotus-miner sectors remove --really-do-it <sectorId>
CommitFailed: lotus-miner sectors update-state --really-do-it <sectorId> Committing
Troubleshoot and restore
the running status of scheduled task scripts check
ps -ef | grep programname
Simple backup methods are used for small-scale mining, one uses inotifywait+rsync to achieve real-time incremental backup, and the other can use scheduled tasks+rsync to achieve backups at intervals of x hours (equivalent to retaining a snapshot of x hours ago) .