Installation Best Practices

AI Tools

Software Recommendations

Note

Refer to the General System Requirements article to view supported operating systems and kernels for the Lightbits cluster software.

Recommendations Validation

Parameters	Recommendation	Description
Bios version	Latest	Lightbits recommends keeping the server BIOS up to date. See the vendor media for additional information.
Lightbox version	Latest	Lightbits recommends keeping Lightbits software up to date, to benefit from the new functionality and improved stability. Use the “lbcli version” command to check your installed version.
Boot drive size	Minimum of 128 GB	Lightbits’ recommendation is to set a separate partition for etcd.
SSD FW version	Latest	Lightbits recommends keeping the drive’s firmware up to date. See the vendor media for additional information. The “nvme list” command displays the current firmware for each installed drive.

NTP Sync

Having the Lightbits cluster nodes stay time synced is crucial to the cluster’s functionality, Lightbits provides events (lbcli list events), for time drifts between servers. If a time drift has been detected, you will need to force the NTP client sync. The following text describes how to do that.

For Chrony:

# chronyc -a 'burst 4/4'

# chronyc -a makestep

For NTP:

# service ntpd stop

# ntpd -gq

# service ntpd start

For getting the most accurate command based on the distro running or latest changes on the NTP client, refer to the relevant ntpd/chronyd pages.

Periodic System Health Check

It is recommended to periodically run the following set of commands:

Parameter	Expected Value
`lbcli list nodes`	All nodes are active.
`lbcli list server`	All servers are in “NoRiskOfServiceLoss” state.
`lbcli list volumes`	All volumes are “FullyProtected”.
`lbcli list events`	No active critical events.
NTP/Chrony config (`cat/etc/ntpd.conf`)	All nodes are synced with the same NTP server and have no time drifts.
Utilized storage (`lbcli get cluster -o yaml`)	The recommended cluster utilization is up to 80%. To calculate the current utilization, use the following formula: `(1-(estimatedFreeLogicalStorage/estimatedLogicalStorage))`

Generating Logs

You can gather logs from a Lightbits server by running the following command from a server:

# lbcli fetch logs --path=logs-`hostname -s`.tar.gz

This will generate a tarball log file for the server into the current working directory. Copy the tarball out using Winscp or SCP to your computer or another storage location. Optionally, you can delete the generated tarball.

To gather the logs from all members of the cluster, repeat the command from the other servers.

Hardware Recommendations

The following table defines the major hardware requirements and guidelines for Lightbits storage nodes.

Requirement	Guideline
CPU	32 cores per Lightbits instance
RAM	Each Lightbits instance requires a minimum of 128 GB of RAM. For precise memory sizing based on specific capacity requirements, consult with Lightbits.
Networking	IPMI: Should be connected Access NIC: 1GbE Data NIC: 25GbE or higher (100GbE recommended)
Boot Device	2x SATA or SSD in RAID1
etcd volume	The drive or volume hosting etcd should support at least 10K IOPs at 4K blocksize. Any healthy SSD should provide this requirement. Note that ETCD can reside on LVM or on the boot device.
Data NVMe Storage Devices	Six or more drives (capacity based on storage requirements),
Journaling NVMe Storage Devices	One or more drives (for better performance, it is recommended to have at least two - up to a maximum of four).

Lightbits storage nodes can have one or two running Lightbits software instances.

Last updated on

Was this page helpful?