On 23/09/15 17:06, Daniel Traynor wrote:
> I've not yet looked deeply into it but i note from
>
> http://fasterdata.es.net/host-tuning/linux/
>
> # recommended for hosts with jumbo frames enabled
> net.ipv4.tcp_mtu_probing=1
>
> and
>
> If you are using Jumbo Frames, we recommend setting tcp_mtu_probing = 1 to help avoid the problem of MTU black holes. Setting it to 2 sometimes causes performance problems.
>
> otherwise i guess it should be 0.
>
> I also note
>
> http://fasterdata.es.net/host-tuning/esnet-perfsonar-tuning/
So if you set
net.ipv4.tcp_mtu_probing = 1
You'll work around brokenness.
Perfsonar is designed to find brokenness, so should be configured to
find it IMHO.
Chris
>
> dan
>
> * Dr Daniel Traynor, Grid cluster system manager
> * Tel +44(0)20 7882 6560, Particle Physics,QMUL
>
> ________________________________________
> From: Testbed Support for GridPP member institutes <[log in to unmask]> on behalf of John Bland <[log in to unmask]>
> Sent: 23 September 2015 16:57
> To: [log in to unmask]
> Subject: Re: mtu probing on perfosnar boxes.
>
> On 23/09/15 16:54, Brian Davies wrote:
>> Hi , I would like to ask….
>>
>> What do other sites have set as the value for:
>>
>> net.ipv4.tcp_mtu_probing
>>
>> in their sysctl on their perfsonar machines.
>>
>> RAL’s are set to zero ( and this seems maybe a mistake on our part?
>
> [root@hepsonar1 ~]# cat /proc/sys/net/ipv4/tcp_mtu_probing
> 0
>
> Certainly not altered by us. Seems to be the default.
>
> John
>
> --
> John Bland [log in to unmask]
> System Administrator office: 220
> High Energy Physics Division tel (int): 42911
> Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2911
> University of Liverpool http://www.liv.ac.uk/physics/hep/
> "I canna change the laws of physics, Captain!"
>
|