Production setup¶

Introduction¶

So you’ve made it past trying out LXD live online, or on a server scavenged from random parts. You like what you see, and now you want to try doing some serious work with LXD.

The vast majority of Linux distributions do not come with optimized kernel settings suitable for the operation of a large number of containers. The instructions in this document cover the most common limits that you’re likely to hit when running containers and suggested updated values.

Common errors that may be encountered¶

Failed to allocate directory watch: Too many open files

<Error> <Error>: Too many open files

failed to open stream: Too many open files in...

neighbour: ndisc_cache: neighbor table overflow!

Server Changes¶

/etc/security/limits.conf¶

Domain	Typ e	Item	Value	Default	Description
*	sof t	nofil e	1048576	unset	maximum number of open files
*	har d	nofil e	1048576	unset	maximum number of open files
root	sof t	nofil e	1048576	unset	maximum number of open files
root	har d	nofil e	1048576	unset	maximum number of open files
*	sof t	memlo ck	unlimited	unset	maximum locked-in-memor y address space (KB)
*	har d	memlo ck	unlimited	unset	maximum locked-in-memor y address space (KB)
root	sof t	memlo ck	unlimited	unset	maximum locked-in-memor y address space (KB) (Only need with `bpf` syscall supervision)
root	har d	memlo ck	unlimited	unset	maximum locked-in-memor y address space (KB) (Only need with `bpf` syscall supervision)

NOTE: For users of the snap, those ulimits are automatically raised by the snap/LXD.

/etc/sysctl.conf¶

Parameter	Value	Default	Description
fs.aio-max-nr	524288	65536	This is the maximum number of concurrent async I/O operations. You might need to increase it further if you have a lot of workloads that use the AIO subsystem (e.g. MySQL)
fs.inotify.max_queued _events	1048576	16384	This specifies an upper limit on the number of events that can be queued to the corresponding inotify instance. 1
fs.inotify.max_user_i nstances	1048576	128	This specifies an upper limit on the number of inotify instances that can be created per real user ID. 1
fs.inotify.max_user_w atches	1048576	8192	This specifies an upper limit on the number of watches that can be created per real user ID. 1
kernel.dmesg_restrict	1	0	This denies container access to the messages in the kernel ring buffer. Please note that this also will deny access to non-root users on the host system.
kernel.keys.maxbytes	2000000	20000	This is the maximum size of the keyring non-root users can use
kernel.keys.maxkeys	2000	200	This is the maximum number of keys a non-root user can use, should be higher than the number of containers
net.core.bpf_jit_limi t	3000000000	264241152	This is a limit on the size of eBPF JIT allocations which is usually set to PAGE_SIZE * 40000. When your kernel is compiled with `CONFIG_BPF_ JIT_ALWAYS_ON =y` then `/proc/sys/n et/core/bpf_j it_enable` is set to `1` and can’t be changed. On such kernels the eBPF JIT compiler will treat failure to JIT compile a bpf program such as a `seccomp` filter as fatal when it would continue on another kernel. On such kernels the limit for eBPF jitted programs needs to be increased siginficantly .
net.ipv4.neigh.defaul t.gc_thresh3	8192	1024	This is the maximum number of entries in ARP table (IPv4). You should increase this if you create over 1024 containers. Otherwise, you will get the error `neighbour: ndisc_cache: neighbor tabl e overflow!` when the ARP table gets full and those containers will not be able to get a network configuration . 2
net.ipv6.neigh.defaul t.gc_thresh3	8192	1024	This is the maximum number of entries in ARP table (IPv6). You should increase this if you plan to create over 1024 containers. Otherwise, you will get the error `neighbour: ndisc_cache: neighbor tabl e overflow!` when the ARP table gets full and those containers will not be able to get a network configuration . 2
vm.max_map_count	262144	65530	This file contains the maximum number of memory map areas a process may have. Memory map areas are used as a side-effect of calling malloc, directly by mmap and mprotect, and also when loading shared libraries.

Then, reboot the server.

Prevent container name leakage¶

Both /sys/kernel/slab and /proc/sched_debug make it easy to list all cgroups on the system and by extension, all containers.

If this is something you’d like to see blocked, make sure you have the following done before any container is started:

chmod 400 /proc/sched_debug
chmod 700 /sys/kernel/slab/

Network Bandwidth Tweaking¶

If you have at least 1GbE NIC on your lxd host with a lot of local activity (container - container connections, or host - container connections), or you have 1GbE or better internet connection on your lxd host it worth play with txqueuelen. These settings work even better with 10GbE NIC.

Server Changes¶

txqueuelen¶

You need to change txqueuelen of your real NIC to 10000 (not sure about the best possible value for you), and change and change lxdbr0 interface txqueuelen to 10000.

In Debian-based distros you can change txqueuelen permanently in
/etc/network/interfaces
You can add for ex.: up ip link set eth0 txqueuelen 10000 to your
interface configuration to set txqueuelen value on boot.
You could set it txqueuelen temporary (for test purpose) with
ifconfig <interface> txqueuelen 10000

/etc/sysctl.conf¶

You also need to increase net.core.netdev_max_backlog value.

You can add net.core.netdev_max_backlog = 182757 to /etc/sysctl.conf to set it permanently (after reboot) You set netdev_max_backlog temporary (for test purpose) with echo 182757 > /proc/sys/net/core/netdev_max_backlog Note: You can find this value too high, most people prefer set netdev_max_backlog = net.ipv4.tcp_mem min. value. For example I use this values net.ipv4.tcp_mem = 182757 243679 365514

Containers changes¶

You also need to change txqueuelen value for all you ethernet
interfaces in containers.
In Debian-based distros you can change txqueuelen permanently in
/etc/network/interfaces
You can add for ex.: up ip link set eth0 txqueuelen 10000 to your
interface configuration to set txqueuelen value on boot.

Notes regarding this change¶

10000 txqueuelen value commonly used with 10GbE NICs. Basically small txqueuelen values used with slow devices with a high latency, and higher with devices with low latency. I personally have like 3-5% improvement with these settings for local (host with container, container vs container) and internet connections. Good thing about txqueuelen value tweak, the more containers you use, the more you can be can benefit from this tweak. And you can always temporary set this values and check this tweak in your environment without lxd host reboot.