perfSONAR User Q&A and Other Discussion

[Andrew Lake <>]

Hi,

That looks like Debian and in that case the packages are only in the “snapshot” repository which is even less stable than staging. Your must have setup your host to be pointing at that repository. Is this not what you wanted?

In order to prevent a stream of these emails, as a general statement, if you have a host pointing at one of our snapshot/nightly or staging repos you will get bleeding edge stuff that might break your system. This is a practice and these are terms that are common in the software community, especially the open source community. We also advertise this in bold letters when we announce RCs, which is the only time we even share instructions for how to point at our staging repo (I don’t think we ever suggest non-developers point at the nightly and snapshot repos). If you are pointing at one of these repos, at some point in the past you made a conscience decision to do so, and as a result you get what you signed-up for. 

We are extremely fortunate to have a small community that has dedicated a few non-production boxes to testing the point at these repos with auto-updates, and they have proved extremely valuable before we officially go live with updates. In the coming weeks we will be seeking even more people to do so as we open up our release candidate testing and move toward the next major release of perfSONAR. 

Thanks,

Andy

On September 15, 2016 at 10:23:29 AM, John Simpkins () wrote:

Same!

Unpacking perfsonar-meshconfig-shared (4.0~2.rc1-1~bpo70+1) over (3.5.1.2-1~bpo70+1) ...

Preparing to unpack .../perfsonar-meshconfig-agent_4.0~2.rc1-1~bpo70+1_all.deb ...

Unpacking perfsonar-meshconfig-agent (4.0~2.rc1-1~bpo70+1) over (3.5.1.2-1~bpo70+1) ...

On Thu, Sep 15, 2016 at 8:55 AM Andrew Lake <> wrote:

Hi,

You didn't miss anything, we didn’t announce the RC yet. We’ll likely officially announce it next week. You must be pointing at Internet2-staging which is our testing repository. Your hosts are part of the WLCG/OSG testbed, correct? We pushed some RPMs to the staging repo so we could get the RPMs to a few well-known places prior to the wider announcement. The problem you encountered on the one host with pscheduler should be fixed next auto-update I think (I still need to push another set of RPMs). There was a problem with one of the pscheduler RPMs disabling itself on upgrade (basically it was disabling itself in the uninstall step without checking if it was a true uninstall or the uninstall of an old package so a new one could be installed as part of an update). Hoping to correct that shortly. Thanks for checking though and sorry for the surprise. 

Thanks,

Andy

On September 14, 2016 at 7:06:06 PM, Michael Petry () wrote:

I noticed that two of my systems that are set for autoupdate picked up toolkit4.0 last night.  Did I miss an announcement?

Mike

P.S. One system seems to run fine, but the other is has pscheduler disabled and error loading test results.