@DrC Thanks for the report. Yes this is a know issue where for a while the kernel corrrectly reported in lower case, earlier CentOS used by Rockstor. Then on later CentOS versions and I think early Leap 15 there was a period where the kernel moved to incorrectly reporting the uuid as uppercase. It then, in Leap 15.2 moved back to properly reporting uuid (read procuct_id, used in Appliance ID) in lowercase.
See the following issue where we standardised to lowercase which is compliant with the uuid standard as per what we now expect (again) form the kernel:
and it’s associated pull request:
As always the problem has a few more angles than are initially obvious. If we allow both cases then we have 2 valid machines per subscription. Plus we have, as you see, the case of matching/catching fake product uuid’s from manufactures.
At least now, from the issue referenced kernel changes we can be assured that the problematic period of uppercase is now over and we have again returned to the ‘proper’ lowercase scenario. Another element in this is that our earlier subscribers will not face this problem as they registered with kernels that correctly reported as lower case. So this only affects our later subscribers for a period of a year or two I think.
All a little tricky, but if you read the issue text and pull request you can see the direction we took. Essentially returning to what was correct when we first set-out on the subscriptions, which was a few years ago now.
So yes, we have a known caveate that for a time some subscriptions were incorrectly upper case and if those folks then upgrade to Rockstor 4 they have to correct the case within Appman.
We could probably do with a doc issue to explain that cases sensitivity is a thing in Appman and if one is using an uppercase Appman Appliance ID it needs to be lower cased, as you have done.
Thanks again for the report. Much appreciated. And well done for finding the solution via Appman. We should probably have the Stable subscription dialog within the Rockstor Web-UI state case is sensitive and to check this within Appman if a code fails to active a new Rockstor 4 subscription.
All would be fine however if the subscription was established with an earlier Rockstor 3 (we have many lower case examples from that period), but later Rockstor 3’s suffer from this situation. So 6 of one and half a dozen of the other was what I was faced with when addressing this issue. Along of course with sorting out the recognition of the fake product uuids.
Bit by bit.
Hope that helps to explain our situation in more detail on that one.