Thread (7 messages) 7 messages, 2 authors, 2002-12-02

Re: NFS queue

From: Ramon Kagan <hidden>
Date: 2002-12-02 17:16:50

I'll look into 2.4.20, as for the hidden patch, it has no affect.  I
checked that out right at the beginning.  I thought originally that it was
the problem.

Ramon Kagan
York University, Computing and Network Services
Unix Team -  Intermediate System Administrator
(416)736-2100 #20263
rkagan@yorku.ca

-------------------------------------
I have not failed.  I have just
found 10,000 ways that don't work.
	- Thomas Edison
-------------------------------------

On Mon, 2 Dec 2002, Trond Myklebust wrote:
On Monday 02 December 2002 14:19, Ramon Kagan wrote:
quoted
quoted
If you are seeing 30Mbit/s on a Gigabit connection, then my vote still
goes to packet loss. I'm seeing ~25MByte/s on writes against a filer
when using 2.4.19 (and about 40MByte/s on reads).
I'm only using 30 Mb/s because that is what the service currently needs,
it's not a limitation.  I was able to create a 250Mb/s load through the
card to the filer.  Would I not get the same behaviour if the card was the
problem?
Depends entirely on how you did your test. If you used UDP with 32k message
sizes, then the 2 should indeed be comparable. If not, ....
quoted
Kernel is 2.4.18 and .config is attached.  Note that I've added LVS
group's hidden patch for direct routing (DR) load balancing.
Try 2.4.20. It has a much improved UDP congestion control.

No idea what your load balancing patch will do, but I suggest that you compare
with and without.

Cheers,
  Trond
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help