FAQ
Why hasn't disk usage changed after deleting an instance?
This is because the storage pool uses sparse storage by default, meaning a large "file block" (loop file) is created initially for storing data, and the storage pool size grows dynamically. New instances will occupy this storage pool, and after deleting an instance, the corresponding usage is marked as available (deleted). When creating a new instance, it can overwrite the corresponding space. However, from the file system's perspective, it seems that space has not been freed, which helps maintain higher performance.
Why are the disk size and storage size inconsistent?
The disk size shown on the web page for an instance is the instance quota, not the actual disk size or the storage pool size.
Why does the free -h
command show different results compared to the panel?
The high memory usage shown on the website includes cache. From the host machine's perspective, the instance uses this much memory, but the cache can be released. The Linux philosophy is to use as much cache as possible to improve performance.
Are there any special considerations for Oracle Cloud?
If you want to test running LXC on Oracle Cloud, pay attention to the following details:
- Disable the firewall, including security group restrictions on the web, and any blocking rules within the host machine (use
iptables -I INPUT -j ACCEPT
). - The Incus server's local IP must be filled in because Oracle Cloud's local IP is different from the public IP.
What to do if there are conflicts between Docker and Incus service's iptables forwarding rules?
You can refer to this issue.
Should I be aware of the provider's TOS when splitting instances from a server?
Yes, you should thoroughly read and comply with the provider's TOS, as some providers restrict nested virtualization.
Regarding performance, if you install Docker on a server, the consumption is almost the same as opening LXC instances with NodeHatch. In fact, LXC sometimes performs even better.
Using a very low-spec KVM to create additional KVM instances is not advisable because the machine will be under high load continuously, which might be considered CPU abuse by the provider. For more information, please refer to Instance Creation Limits.
Some VPS providers do not care whether you use LXC, KVM, or Docker; the concern is not about performance loss but about the increased risk of abuse when you split the machine for many other users. If you are just using it for your own convenience to manage services and provide basic environments, it generally does not violate the provider's TOS.