Initializing

Overview

The fdctl configure command is used to setup the host environment so Firedancer can run correctly. It should be run each and every time before the validator is started, and also should be rerun if the configuration file changes, even if the validator has not yet been started.

WARNING

Configuring the system for Firedancer has many moving parts and you should prefer to run fdctl configure init all if possible. The stages are described here in detail for advanced use cases and operators only.

The configure command is run like fdctl configure <mode> <stage>... where mode is one of:

• init Configures the provided stages if they are not already configured.
• check Check if each stage is already configured. The command will exit with an error code if they are not. check never requires privileges and will not make any changes to the system.
• fini Unconfigure (reverse) the stage if it is reversible.

stage can be one or more of hugetlbfs, sysctl, xdp, xdp-leftover, ethtool, workspace-leftover, or workspace and these stages are described below. You can also use the stage all which will configure everything.

Stages have different privilege requirements, which you can see by trying to run the stage without privileges. The check mode never requires privileges, and the init mode will only require privileges if it needs to actually change something.

hugetlbfs

The hugetlbfs stage is used to reserve huge (2MiB) and gigantic (1GiB) memory pages from the Linux kernel for use by Firedancer. See also the kernel documentation of these pages. Almost all memory in Firedancer is allocated out of these pages for performance reasons.

This is a two step process. First, the number of huge and gigantic pages available on the entire system is increased in the kernel by increasing /sys/devices/system/node/node0/hugepages/hugepages-1048576kB/nr_hugepages until the free_hugepages value is high enough for all the memory needs of the validator.

Once the pages have been reserved globally in the kernel pool, they are assigned specifically to Firedancer by creating a hugetlbfs mount at each of /mnt/.fd/.gigantic/ and /mnt/.fd/.huge for gigantic and huge pages respectively. These paths can be configured in the TOML file under the [hugetlbfs] section. Lets run it:

$ sudo fdctl configure init hugetlbfs
NOTICE  hugetlbfs ... unconfigured ... mounts `/mnt/.fd/.huge` and `/mnt/.fd/.gigantic` do not exist
NOTICE  hugetlbfs ... configuring

$ cat /proc/mounts
none /mnt/.fd/.gigantic hugetlbfs rw,seclabel,relatime,pagesize=1024M,min_size=540092137472 0 0
none /mnt/.fd/.huge hugetlbfs rw,seclabel,relatime,pagesize=2M,min_size=95124124 0 0

This stage requires root privileges, and cannot be performed with capabilities. If the required hugetlbfs mounts are already present, with at least the amount of memory reserved that we required then the init mode does nothing and the check mode will return successfully without requiring privileges.

The fini mode will unmount the two filesystems, and remove them from /mnt/.fd/, although it will leave the /mnt/.fd/ directory in place. The fini mode will not succeed if memory from the mounts is mapped into a running process. This will return the huge and gigantic pages that Firedancer had reserved to the global kernel pool, although we will not decrease the global pool size, even if it was earlier increased during init.

TIP

The hugetlbfs step should be run immediately when the system is booted. If run later, it may fail because the operating system memory is fragmented and a large contiguous block cannot be reserved.

sysctl

It is suggested to run Firedancer with certain kernel parameters tuned for best performance. The sysctl stage will check and configure these parameters. The stage will only increase values to meet the minimum, and will not decrease them if the minimum is already met.

Sysctl	Minimum	Description
/proc/sys/net/core/rmem_max	134217728	Solana Labs network performance tuning.
/proc/sys/net/core/rmem_default	134217728	Solana Labs network performance tuning.
/proc/sys/net/core/wmem_max	134217728	Solana Labs network performance tuning.
/proc/sys/net/core/wmem_default	134217728	Solana Labs network performance tuning.
/proc/sys/vm/max_map_count	1000000	Solana Labs accounts database requires mapping many files.
/proc/sys/fs/file-max	1024000	Solana Labs accounts database requires opening many files.
/proc/sys/fs/nr_open	1024000	Solana Labs accounts database requires opening many files.
/proc/sys/net/core/bpf_jit_enable	1	Firedancer uses BPF for kernel bypass networking. BPF JIT makes this faster.
/proc/sys/net/ipv4/conf/lo/rp_filter	2	Enable loose mode for reverse path filtering on the loopback interface. Loose mode is required for the XSK socket to successfully send packets to loopback.
/proc/sys/net/ipv4/conf/lo/accept_local	1	Accept packets with local source addresses on the loopback interface. This is required for the XSK socket to successfully send packets to loopback.

The init mode requires either root privileges, or to be run with CAP_SYS_ADMIN. The fini mode does nothing and kernel parameters will never be reduced as a result of running configure.

xdp

Firedancer uses XDP (express data path), a Linux feature for doing high performance kernel bypass networking. For more background see the kernel documentation.

To configure XDP, a BPF program is loaded onto both the configured network interface [tiles.net.interface] and the loopback interface lo. This BPF program intercepts packets matching a Firedancer listen port before they reach the kernel. Matching packets are routed directly to Firedancer.

WARNING

Packets intercepted by the BPF program will not appear under standard network monitoring tools like tcpdump.

The BPF program is loaded into /sys/fs/bpf/<name>/ and will remain loaded until fini is run. If loaded, packets for the target ports will be intercepted even when Firedancer itself is not running. fini will fully unload the program, packets for the target ports will resume being routed to the kernel and regular networking stack.

This stage must be run, and it is not possible to manually configure it. The stage not only loads the program, but sets up special configuration objects (BPF maps) so that it functions correctly.

The stage must be rerun any time the system is rebooted, any time Firedancer is updated, or any time the configuration file changes. The init mode requires root or both CAP_SYS_ADMIN and CAP_NET_RAW.

ethtool

In addition to XDP, Firedancer uses receive side scaling (RSS) to improve network performance. This uses functionality of modern NICs to steer packets to different queues to distribute processing among CPUs. See the kernel documentation for more information.

In Firedancer, each net tile serves one network queue, so the ethtool stage will modify the combined channel count of the configured network device [tiles.net.interface] to be the same as the number of net tiles, [layout.net_tile_count]. If your NIC does not support the required number of queues, you will need to reduce the number of net tiles, potentially down to one for NICs which don't support queues at all.

The command run by the stage is similar to running ethtool --set-channels <device> combined <N> but it also supports bonded devices. We can check that it worked:

$ sudo fdctl configure init ethtool
NOTICE  ethtool ... unconfigured ... device `ens3f0` does not have right number of channels, got 2, expected 4
NOTICE  ethtool ... configuring

$ ethtool --show-channels ens3f0
Channel parameters for ens3f0:
Pre-set maximums:
RX:		64
TX:		64
Other:		1
Combined:	64
Current hardware settings:
RX:		0
TX:		0
Other:		1
Combined:	4

The stage only needs to be run once after boot but before running Firedancer. It has no dependencies on any other stage, although it is dependent on the number of in your configuration.

Changing device settings with ethtool requires root privileges, and cannot be performed with capabilities.

workspace

A Firedancer workspace is a special in-memory file with extension .wksp created in the shmem mountpoints described above. It can be thought of as just a chunk of memory. Almost all memory in Firedancer is reserved, allocated, and initialized before running the program. The allocations are made in this workspace, which has a special header so we can look up where each one is.

In the future, Firedancer will support persistent workspaces that are kept between runs. If that were supported, you could restart the program very quickly from existing memory.

For now though, the stage must be run every time before running Firedancer to get a cleanly initialized memory space.

The workspace stage only initializes memory that was already reserved in a large block by the shmem stage, and does not require any privileges. It needs to be run after the shmem stage but does not have other dependencies.

The check mode will always fail, as the workspace always needs to be reinitialized. The fini mode will remove existing workspaces from the shared memory mount points.

TIP

It is possible to run Firedancer without rerunning most of the stages of fdctl configure, except workspace, if you are sure that your environment has been set up correctly.