Increase default NFSv4 server size "max_block_size" to 4MB

Increase default NFSv4 server size "max_block_size" to 4MB

[Cedric Blancher]

Good morning!

As debated a while ago, can the default NFSv4 server size for
"max_block_size" be increased to 4MB, please?

Ced
-- 
Cedric Blancher <cedric.blancher@gmail.com>
[https://plus.google.com/u/0/+CedricBlancher/]
Institute Pasteur

Re: Increase default NFSv4 server size "max_block_size" to 4MB

[Chuck Lever]

On Mon, Mar 16, 2026, at 3:51 AM, Cedric Blancher wrote:
> As debated a while ago, can the default NFSv4 server size for
> "max_block_size" be increased to 4MB, please?

There is an administrative setting to raise this limit for
recent versions of the kernel. Can you report your experience
when you raise the limit? Hiccups, performance issues, etc? I
would kind of like this exercise to be data-driven.

What is still unknown to me is which NFS client implementations
can support 4MB or 8MB. Without client support, an increase in
the default in NFSD doesn't mean anything. Rick, Anna, Roland?

-- 
Chuck Lever

Re: Increase default NFSv4 server size "max_block_size" to 4MB

[Roland Mainz]

On Mon, Mar 16, 2026 at 1:39 PM Chuck Lever <cel@kernel.org> wrote:
> On Mon, Mar 16, 2026, at 3:51 AM, Cedric Blancher wrote:
> > As debated a while ago, can the default NFSv4 server size for
> > "max_block_size" be increased to 4MB, please?
>
> There is an administrative setting to raise this limit for
> recent versions of the kernel. Can you report your experience
> when you raise the limit? Hiccups, performance issues, etc? I
> would kind of like this exercise to be data-driven.
>
> What is still unknown to me is which NFS client implementations
> can support 4MB or 8MB. Without client support, an increase in
> the default in NFSD doesn't mean anything. Rick, Anna, Roland?

Linux 6.17 NFS server:
$ service nfs-server stop
$ printf '%d\n' $((4*1024*1024)) >/proc/fs/nfsd/max_block_size
$ service nfs-server start

Mounting the filesystem with ms-nfs41-client and compiling Cygwin
bash.exe and ms-nfs41-client with Visual Studio etc. works fine...
... and I verified that |FATTR4_MAXREAD| and |FATTR4_MAXWRITE| have
the value |4194304| as expected.

I didn't make any benchmarks yet, but quick observation shows that
cloning a git from a git bundle on the same NFSv4.2 filesystem is up
from 8.1MB/s to 8.8MB/s.

----

Bye,
Roland
-- 
  __ .  . __
 (o.\ \/ /.o) roland.mainz@nrubsig.org
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 641 3992797
 (;O/ \/ \O;)

Re: Increase default NFSv4 server size "max_block_size" to 4MB

[Rick Macklem]

On Mon, Mar 16, 2026 at 5:41 AM Chuck Lever <cel@kernel.org> wrote:
>
>
> On Mon, Mar 16, 2026, at 3:51 AM, Cedric Blancher wrote:
> > As debated a while ago, can the default NFSv4 server size for
> > "max_block_size" be increased to 4MB, please?
>
> There is an administrative setting to raise this limit for
> recent versions of the kernel. Can you report your experience
> when you raise the limit? Hiccups, performance issues, etc? I
> would kind of like this exercise to be data-driven.
>
> What is still unknown to me is which NFS client implementations
> can support 4MB or 8MB. Without client support, an increase in
> the default in NFSD doesn't mean anything. Rick, Anna, Roland?
Although it has not seen much testing, it is possible to do a > 1Mbyte NFSv4
mount in FreeBSD.
For a 2Mbyte mount, (the only size > 1Mbyte I've tried) the settings would be..
In /boot/loader.conf
kern.maxphys=2097152
vfs.maxbcachebuf=2097152

and in /etc/sysctl.conf
kern.ipc.maxsockbuf=9455616

Then a mount will use 2Mbytes if the server supports it.

I doubt anyone does this, but it works for trivial tests.

Maybe I'll do it during the next bakeathon? rick
ps:It would be really nice if Roland could show up
     (you can attend remotely via tailscale like I do)
     at the next bakeathon.

>
> --
> Chuck Lever
>

Re: Increase default NFSv4 server size "max_block_size" to 4MB

[Anna Schumaker]

On Mon, Mar 16, 2026, at 8:39 AM, Chuck Lever wrote:
> On Mon, Mar 16, 2026, at 3:51 AM, Cedric Blancher wrote:
>> As debated a while ago, can the default NFSv4 server size for
>> "max_block_size" be increased to 4MB, please?
>
> There is an administrative setting to raise this limit for
> recent versions of the kernel. Can you report your experience
> when you raise the limit? Hiccups, performance issues, etc? I
> would kind of like this exercise to be data-driven.
>
> What is still unknown to me is which NFS client implementations
> can support 4MB or 8MB. Without client support, an increase in
> the default in NFSD doesn't mean anything. Rick, Anna, Roland?

The NFS client would need a code change to support >1MB sizes. I
spent some time playing around with 4MB yesterday and it passed
all my tests.

Anna

>
> -- 
> Chuck Lever

Re: Increase default NFSv4 server size "max_block_size" to 4MB

[Mike Snitzer]

On Tue, Mar 17, 2026 at 09:51:23AM -0400, Anna Schumaker wrote:
> 
> 
> On Mon, Mar 16, 2026, at 8:39 AM, Chuck Lever wrote:
> > On Mon, Mar 16, 2026, at 3:51 AM, Cedric Blancher wrote:
> >> As debated a while ago, can the default NFSv4 server size for
> >> "max_block_size" be increased to 4MB, please?
> >
> > There is an administrative setting to raise this limit for
> > recent versions of the kernel. Can you report your experience
> > when you raise the limit? Hiccups, performance issues, etc? I
> > would kind of like this exercise to be data-driven.
> >
> > What is still unknown to me is which NFS client implementations
> > can support 4MB or 8MB. Without client support, an increase in
> > the default in NFSD doesn't mean anything. Rick, Anna, Roland?
> 
> The NFS client would need a code change to support >1MB sizes. I
> spent some time playing around with 4MB yesterday and it passed
> all my tests.

Is this something you'd be open to posting to the list?

Would be interested to see how 'git clone' improves with it.

Thanks,
Mike