Reduce sunrpc.tcp_max_slot_table_entries

[donald]

PR #93 did not work as expected.

Further analysis showed, that the parallelism of page io
(which includes buffered reads) is not limitted by nfs.max_session_slots
but by sunrpc.tcp_max_slot_table_entries.

The default value is 65536 (RPC_MAX_SLOT_TABLE_LIMIT).

This parameter can be modified for the running system via
/proc/sys/sunrpc/tcp_max_slot_table_entries and is applied to
new server connections.

Experiment (on simsalabimbambasaladusaladim):

# preload server cache
ssh mrtorgue cat /project/stresser/files/random_10G.dat \> /dev/null

umount /project/stresser
echo 3 > /proc/sys/vm/drop_caches
time cat /project/stresser/files/random_10G.dat > /dev/null

# result: 0m11.429s  ca. 14 nfsd visible active in `top` on server

echo 4 > /proc/sys/sunrpc/tcp_max_slot_table_entries
umount /project/stresser
echo 3 > /proc/sys/vm/drop_caches
time cat /project/stresser/files/random_10G.dat > /dev/null

# 0m12.546s , 4 nfsd visible active in `top` on server

echo 65536 > /proc/sys/sunrpc/tcp_max_slot_table_entries
umount /project/stresser
echo 3 > /proc/sys/vm/drop_caches
time cat /project/stresser/files/random_10G.dat > /dev/null

# result: 0m11.937s , ca. 16 nfsd visible active in `top` on server

Reduce sunrpc.tcp_max_slot_table_entries to 4 instead of nfs.max_session_slots.

[pmenzel]

So why 4?

What resources does sunrpc.tcp_max_slot_table_entries depend on?

[donald]

• As small as possible without causing bad effects. 4 doesn't seem to change user experience very much. See above timing. Also, I've rebooted my workstation with it and don't "feel" a disadvantage. However, we yet need to test a active cluster system with many processes and doing grep xx a>b or somesuch nonsense, too. This is the "need more testing" label.

• I don't understand the other question.

[donald]

So lets start with 8 then?

[wwwutz]

I think 4 is ok, we'll see what happens. Dont forget there might be > 20 clients accessing one server.
The question should not be "why 4" but "hey, my research research revealed 4 is a bad value, please change it. to 8".
Four is a good value.

[donald]

/proc/sys/sunrpc/tcp_max_slot_table_entries not availabe on every kernel. ( CONFIG_SUNRPC_DEBUG probably required, which we don't have on old kernels).

[donald]

https://bugzilla.kernel.org/show_bug.cgi?id=204019