For my case, here is the result why we had this problem.
On my compute node (a X6 with 88 vcpu available), we had 3 OVM running :
- VM1 : 44 vcpu
- VM2 : 34 vcpu
- VM3 : 8 vcpu
- dom0 : 20 vcpu
which gives a total of 106 vcpu needed (on 88 available).
When we downsize the VM2 with 34 vcpu VM to 16 vcpu, it started working because we get a total of 88 vcpu needed on 88 available.
Thank you for your feedback Tim!
It will be definitely useful for people having the same issue.
I will continue investigation for my particular case.
We had almost the exact same issue except we were running on KVM instead of Xen. According to the Oracle DB release notes, UEKR5 is not certified with 12.1. Once we installed UEKR4 everything began working correctly.
Thank you for sharing your experience
I was going to upgrade 12.1 to 19c, but faced those hanging issues during the pre-checks. Database was hanging after even small loads, so I doubted that the upgrade would complete successfully.
Per Doc ID 2551169.1 19c is certified with the following OEL releases:
Oracle Linux 7.4 with the Unbreakable Enterprise Kernel 4: 4.1.12-124.19.2.el7uek.x86_64 or later
Oracle Linux 7.4 with the Unbreakable Enterprise Kernel 5: 4.14.35-1818.1.6.el7uek.x86_64 or later
So I will try to perform the upgrade using OEL 7.4 with UEK 4 as you suggested.
I will share the results later.
I finally upgraded our EBS database from 22.214.171.124 to 126.96.36.199 and then to 19c on the following OEL 7.4 kernel: 4.1.12-188.8.131.52.el7uek.x86_64
So far I didn't notice the hanging issue. So assuming the incorrect kernel version was the reason
Will upgrade one more database soon to be sure, but marking this thread as solved alread
Thanks everyone for help!