perfSONAR User Q&A and Other Discussion

[Andrew Lake <>]

Hi David,

Is this host at an address that is reachable? If so can you share the address? I can look at the pscheduler tasks and get a better idea of what is happening. The tasks only archive results where you tell them to, so they only post to localhost and/or the central archive if that is what you configured them to do. You can look in /etc/perfsonar/meshconfig-agent-tasks.conf at the measurement_archive blocks to see what you have configured.  It looks like pscheduler is having trouble accessing one of the archives which could be caused by a number of things such as a misconfigured URL to the archive on the client, issues on the server, etc. 

Thanks,
Andy

On Mon, Sep 10, 2018 at 12:34 PM, David Szydloski <> wrote:

Hello All,

 

OS: Ubuntu 16.04

perfsonar version: 4.0.2.5-1

 

I am running perfSONAR on Ubuntu boxes in a mesh configuration with one central management node. I have one node currently not able to upload any results to the central management node, showing in /var/log/pscheduler/pscheduler.log that the remote host is getting "400: Invalid JSON returned" when it tries to send results into the central archive.

 

On the central archiver node, there are no errors in /var/log/cassandra/system.log nor in the /var/log/esmond logs. In the apache logs, I only see logs of POST attempts:

 

10.68.8.26 - - [10/Sep/2018:16:23:22 +0000] "POST /esmond/perfsonar/archive/ HTTP/1.1" 201 7044 "-" "python-requests/2.9.1"

10.68.8.26 - - [10/Sep/2018:16:23:22 +0000] "POST /esmond/perfsonar/archive/ HTTP/1.1" 201 7044 "-" "python-requests/2.9.1"

 

Back in March I wrote in about another instance of this issue, https://lists.internet2.edu/sympa/arc/perfsonar-user/2018-03/msg00032.html , and Andy mentioned that when remote nodes are reporting results, the PUT commands are sending data to the central manager's cassandra instance. So issues connecting to the central archive node can sometimes be resolved by restarting cassandra on the central manager instance. I've done that a few times without a change in the behavior. I've also tried reinstalling perfsonar-toolkit 2-3 times on this node in case I messed up the install.

 

My guess based on all of this is that my remote node is not able to authenticate tests results in part because the metadata isn't getting generated correctly on the remote server. Here are my questions:

1. In a mesh configuration with a central management node, is the metadata to use in reporting generated on the client or the central-archiver?
2. What process is trying (and in my case possibly failing) to generate this metadata?
3. In a mesh configuration with a central management node, does the remote host need to interact at all with the local Esmond instance at all?

 

Thanks for your time and assistance,

David Szydloski

--
To unsubscribe from this list: https://lists.internet2.edu/sympa/signoff/perfsonar-user