I recently been experiencing some erratic behavior after tweaking some of the external sites we have enabled for patching. I made the changes in the effort to filter Amazon Web Service’s VM’s we have from our internal clients. Unfortunately, it seems to have caused an issue with the Red Hat 6 site I have. All of a sudden Oracle Linux 6 machines are subscripting to the site even when explicitly tell it not to.
I have since opened a PMR with IBM but was wondering if anyone has ever encountered this issue and resolved it successfully? I have tried tweaking the relevance, clearing the cache, removing and reenabling the site all with no success.
They endpoints are returning as Oracle Linux and I am on the latest 9.5.4.58 build. Strangely, this afternoon the relevance seems to be working again after removing and enabling the site for a 5th time.
You may have to wait after disabling the site. I presume you are moving the subscription to “No Computers” and then moving it back to a custom relevance. You need to wait for the actionsite to propagate to all the endpoints and for the unsubscription to process and then add them again usually.
I tried selecting the radio button to no computers but the most recent attempt involved me disabling the site from the License Overview panel and re-enabling it.
When you turn the site to “No Computers” I presume you did click the “Save Changes” at the top? This makes an actionsite change (so MO’s only can do it) and has to propagate and be processed.
I would say yes, if you are going to disable a site, you should flip it to No Computers for the site subscription first and let that propagate for a while. You can also remove read access to it for non master operators. Once you have done this, it should no longer be having any impact except for a tad on master operator consoles and the root server itself. Then you can eventually disable it, but probably no need to be in a hurry to do so.
Part of the problem might be the fact that Oracle Linux is a flavor of CentOS/RHEL in most cases. Not sure if the site content works for it, but if it does, then it probably should be relevant. I have never dealt with Oracle Linux other than playing with it in docker, so I’m not sure of this.
After reading everyone’s post I believe part of my problem is that I was not waiting long enough for the changes to propagate before re-enabling the site. The issue seems to have resolved itself for now but I agree that the clones of the standard RHEL/CentOS can cause issues at times with regards to relevance.