perfSONAR User Q&A and Other Discussion

[Andrew Lake <>]

Hi,

Comments inline:

On March 4, 2016 at 4:14:44 PM, Liu, Dengfeng (NIH/CIT) [C] () wrote:

Our servers got auto update to 3.5.1 and the changes have caused service interruption to us because we are using a central MA and a customized maddash dashboard configuration. Also many of the puppet modules need changes. 
1. perfsonar changed from /opt to /usr/lib. As convention, we have put most applications under dedicated disk partition /opt, and now it's using same disk as /. 

We apologize for the inconvenience but FWIW almost no data is written or read from the old /opt directory and most of the data applications worked with has always been under /var. I understand you may have had other reasons for wanting to do the split but just making that note.

2, Central MA using IP based authentication seems not working and message shows will be removed in Django 1.9. 

This will continue to work and I believe that is a misinterpretation of some of the messages (which are admittedly confusing) you see at upgrade. There are some warnings about some of the ip fields the SNMP collection portions of esmond uses, which is not utilized by perfSONAR nor related to authentication. The IP authentication used to store data should continue to work as before and there are currently no plans to remove it.

3, testing nodes have problem posting data to the central MA. 

You need to upgrade your testing nodes as well. Specifically perfsonar-regulartesting (formerly perl-perfSONAR_PS-RegularTesting) needs to be updated to 3.5.1 once your MA is on esmond 2.0. Hopefully that resolves your problkems. Please let us know if it doesn’t.

Anyone has any experiences? 
Thanks