Bug 219645 - max NFS I/O size is not tunable
Summary: max NFS I/O size is not tunable
Status: New
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: CURRENT
Hardware: Any Any
: --- Affects Only Me
Assignee: Rick Macklem
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-05-29 22:51 UTC by Rick Macklem
Modified: 2017-06-02 21:19 UTC (History)
0 users

See Also:


Attachments
make max NFS client I/O size tunable (11.38 KB, patch)
2017-05-29 22:51 UTC, Rick Macklem
no flags Details | Diff
make max NFS client I/O size tunable (12.39 KB, patch)
2017-05-30 12:16 UTC, Rick Macklem
no flags Details | Diff
make max NFS client I/O size tunable (9.57 KB, patch)
2017-06-02 21:19 UTC, Rick Macklem
no flags Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Rick Macklem freebsd_committer 2017-05-29 22:51:08 UTC
Created attachment 183045 [details]
make max NFS client I/O size tunable

The maximum size of an I/O operation done by the NFS client is the
largest buffer cache block size (MAXBCACHEBUF).
This can only be changed by recompiling the kernel.

This patch make it a tunable called vfs.maxbcachebuf and tweaks the
NFS client to use this size for nm_rsize/nm_wsize by default.

It also generates a console log message if kern.ipc.maxsockbuf needs
to be increased.
Comment 1 Rick Macklem freebsd_committer 2017-05-30 12:16:04 UTC
Created attachment 183067 [details]
make max NFS client I/O size tunable

Patch updated slightly so that it generates better warning messages
related to kern.ipc.maxsockbuf.
Comment 2 Rick Macklem freebsd_committer 2017-06-02 21:19:00 UTC
Created attachment 183164 [details]
make max NFS client I/O size tunable

This is a simplified patch that does not tune BKVASIZE, since
it is not apparent that this should be done at this time.
(Both kib@ and bde@ have indicated that they do not think it is
 necessary.)