Skip to Content.
Sympa Menu

perfsonar-user - Re: [perfsonar-user] OWAMP tests not scheduling reliably in mesh

Subject: perfSONAR User Q&A and Other Discussion

List archive

Re: [perfsonar-user] OWAMP tests not scheduling reliably in mesh


Chronological Thread 
  • From: Mark Feit <>
  • To: "Dale W. Carder" <>
  • Cc: Casey Russell <>, "" <>
  • Subject: Re: [perfsonar-user] OWAMP tests not scheduling reliably in mesh
  • Date: Fri, 23 Mar 2018 19:40:33 +0000
  • Accept-language: en-US
  • Authentication-results: es.net; dkim=none (message not signed) header.d=none;es.net; dmarc=none action=none header.from=internet2.edu;
  • Ironport-phdr: 9a23:/0GPiBw5IuFPLKPXCy+O+j09IxM/srCxBDY+r6Qd2ukVIJqq85mqBkHD//Il1AaPAd2Araocw8Pt8InYEVQa5piAtH1QOLdtbDQizfssogo7HcSeAlf6JvO5JwYzHcBFSUM3tyrjaRsdF8nxfUDdrWOv5jAOBBr/KRB1JuPoEYLOksi7ze+/94HdbglSmDaxfa55IQmrownWqsQYm5ZpJLwryhvOrHtIeuBWyn1tKFmOgRvy5dq+8YB6/ShItP0v68BPUaPhf6QlVrNYFygpM3o05MLwqxbOSxaE62YGXWUXlhpIBBXF7A3/U5zsvCb2qvZx1S+HNsDtU7s6RSqt4LtqSB/wiScIKTg58H3MisdtiK5XuQ+tqwBjz4LRZoyeKfhwcb7Hfd4CRWRPQNtfVzBPDI2/YYsADesBMvpXoYbyvFYOsQCxBQywCO/z1jNFhHn71rA63eQ7FgHG2RQtEdUUv3TOr9X+KaccXv60zKnU0D7OaPZW2Tnj54jJbhAsuvCMXbRsccXP00kvER3KjkmOpYD/ITyay/kNvnGd4uF9Vuyvk3Yqpx9rrTSz28shj5TFip8Ixl3B+yV0z5o5KcG9RUN5f9GpH4dcuiSfOodoXs8vQnlktDsmxrAFo5K3YisHxI45yxPQb/GLaZWE7g/+WOqJLzd3mnFodK66ihu38UWv1vHwWtex3VtKtCVIlsfDu3UJ2hHd98eKSf5w8Vm81jmV2QDe7PxPL1oumqrBMZEhx6Y9lpoNvkTHGS/7gF34gbOReEk45+Sk8vnpbK37qpOFMI97kR/xPr4pmsyiHeQ3KQ8OX3Wd+euhzrHj5Vf5QLJWjvIojqbZrJHaJcMdpqKjBA9Vz5oj6xK4Dzeh09QUh2UILFVAeB6fjojpPU/BIOzgAPuhmVugjCpnyvXbMrH8H5nAK3bDnbj9cbph7kNRyhQ8wN9D659RD7wNPu7/V07puNzdFBA5Mgi0w+j9CNV604MTQWePAqGHP6PTrV+F/eMvI++QZIALojb9LeYq5+LwgXMjh1ASYLSp0YMNaH+kBvRmP1mZYX30j9cZD2gKpAQ+TPftiF2ESzFTfW++X78n5j4lEoKrFoPDRoG2gLydxye3AIdaZmFAClCQD3jobYOEVOkQaC6MOMNujCELVaXyA7MmgFuCvQv3g4UhZsHd4CEVr9irgNRq7OTJmRwo3TFvScKQzzfeYXtzmzYtTiU1lI5yoFA1nlKN3Klkq/1eCdFJ4f5VCEE3OYOKnL8yMMz7Rg+UJoTBc12hWNjzRGhpFt8=
  • Spamdiagnosticoutput: 1:0

Dale W. Carder writes:

Thus spake Mark Feit
()
on Fri, Mar 23, 2018 at 05:35:48PM +0000:
>
> That shouldn’t cause any problems. All of pScheduler’s interaction is
based around URLs, which don’t have a way to specify protocol. The way
getaddrinfo() is configured on dual-stack systems will make the pSchedulers
talk amongst themselves with IPv4 even if the tests specify IPv6.

I'm curious, what was the rationale in implementing the opposite of what
is typically the default address selection order?

Sorry, I spake backwards. The host names in pScheduler URLs are ultimately
passed down to the system's getaddrinfo(), which will return IPv6 first if it
makes sense to do so. We use a Python library called Requests to handle all
of the HTTP(S) work.

The negotiation process isn't tied to the test parameters other than in one
spot where the test plugin gets asked to return a list of the participants'
host names. There's no protocol associated with those, so it's possible that
negotiation for a test that specifies one protocol could be done with the
other if things fall out the right way.

--Mark





Archive powered by MHonArc 2.6.19.

Top of Page