Hi Chris
The hosts down the left hand side are the sources. At the intersection
of each pair the top square represents results from the source archive
and the bottom square the destination archive. For the line where RALPP
is the source almost all the top boxes are orange indicating a problem
with the archive on your host. So the two measures of 4.047 Gbps and
0.04 Gbps come from the ECDF host:
http://gridpp-ps-band.ecdf.ed.ac.uk/toolkit/
The result you quote of 5.91 Gbps whilst we can see it directly from the
toolkit on your host
http://heplnx130.pp.rl.ac.uk/toolkit/
it is not visible on maddash because of the issue with reading it from
the archive on your host. You'll notice some hints here on what might be
wrong
https://maddash.aglt2.org/maddash-webui/details.cgi?uri=/maddash/grids/UK+Config+-+IPv4+Bandwidth+Tests/heplnx130.pp.rl.ac.uk/gridpp-ps-band.ecdf.ed.ac.uk/Throughput
A majority (but not all) of tests initiated by this site are failing in
both incoming and outgoing directions
Category:CONFIGURATION
Potential Solutions:
Check if the sites that are failing are blocking TCP port 4823.
Verify that /usr/lib/perfsonar/bin/generate_configuration doesn't
throw any errors.
Verify that /etc/perfsonar/regulartesting.conf contains the proper
tests
Restart perfsonar-regulartesting, it may not have picked-up
configuration changes (/etc/init.d/perfsonar-regulartesting restart)
Duncan
On 25/10/16 14:50, Chris Brew wrote:
> Hi Duncan,
>
> I've not done an exhaustive check but the numbers on that page don't seem to agree with the numbers I see on the web interfaces.
>
> For instance on transfers between RALPP and ECDF, the dashboard has either unavailable or 0.04GB/s, whereas the toolkit page has 5.91Gb/s, and on the packet loss side the highest I see is about 2.5e-3.
>
> Is the page not getting proper responses from my servers?
>
> Yours,
> Chris.
>
> Sent from my iPad
>
>> On 24 Oct 2016, at 19:18, Duncan Rand <[log in to unmask]> wrote:
>>
>> On the subject of perfSONAR I have re-created the old UK mesh and it includes testing over IPv6:
>>
>> http://maddash.aglt2.org/maddash-webui/index.cgi?dashboard=UK%20Config
>>
>> It is showing that currently both the RAL latency hosts are uncontactable, the Durham and Oxford hosts are inaccessible over IPv6 (I’ve submitted tickets about that). We’re missing Glasgow. Sheffield and RHUL are not working well and there is significant packet loss inbound to RALPP and outbound from Bristol.
>>
>> The good news is that we now have 8 dual-stack perfsonar hosts with Liverpool coming online recently and Lancaster almost there. It would be good to fix all these issues now we have a mesh view confined to the UK.
>>
>> Duncan
>>
>>
>>> On 24 Oct 2016, at 16:09, Matt Doidge <[log in to unmask]> wrote:
>>>
>>> Heya all,
>>> We're still up to our beards (real or metaphorical) in tickets here in the UK this week - 32 of the blighters. I suspect many of us have been busy sorting out Dirty Cow issues (which will be discussed tomorrow in the meeting - please tune in tomorrow to find out more from the Ian on this, but there *will* be obligatory kernel updates in the near future).
>>>
>>> Using that excuse this is another weak ticket update. I'll do a proper job next week, promise,
>>>
>>> 32 Open UK Tickets
>>>
>>> Looks like they can be closed:
>>> https://ggus.eu/?mode=ticket_info&ticket_id=123504 (20/9) - Tier 1 T2K ticket, with T2K quiet on the grid and Jon P's departure I would close this ticket.
>>>
>>> https://ggus.eu/?mode=ticket_info&ticket_id=124411 (12/10) - QMUL ticket regarding IceCube GPU glideins. Looks like things are working and the ticket can be closed, interesting stuff though.
>>>
>>> Need an Update:
>>> https://ggus.eu/?mode=ticket_info&ticket_id=122378 (4/10) - Glasgow Perfsonar ticket. I'm being a bit unfair singling this one out though.
>>>
>>> https://ggus.eu/?mode=ticket_info&ticket_id=123870 (14/9) - Manchester Perfsonar ticket.
>>>
>>> https://ggus.eu/?mode=ticket_info&ticket_id=121687 (10/10) - Tier 1 Perfsonar ticket - a reminder to update how do things look after a fortnight of statistics with the new router.
>>>
>>> https://ggus.eu/?mode=ticket_info&ticket_id=123962 (13/10) - Liverpool biomed ticket, after being taught the wonders of space tokens by John I noticed that Sorina's update contained a question that could do with a response.
>>>
>>> https://ggus.eu/?mode=ticket_info&ticket_id=124244 (5/10) - Tier 1 Ticket from LHCB for a cvmfs issue at RAL. Is this still a problem? No update for 19 days.
>>>
>>> https://ggus.eu/?mode=ticket_info&ticket_id=124431 (14/10) - Tier 1 ticket from atlas regarding a frontier squid. Checking the plots things still don't look right (at least to my eyes).
>>>
>>> Cheers all,
>>> Matt
>>
|