python3 error and unschedule reboot

I can not proceed without some definite vector. Since the target system was migrated to SCALE and is OK there it does not sound like we get any more information. Closing the ticket.

You’ve practically replaced all the software aside of ZFS. So I am not exactly surprised that it may solve the issue, if we accept that the memory corruption is caused by software. Since you now mentioned you are using NFS, it may be some NFS-specific issue, but that is only a small step in the diagnostic process. We generally recommend using iSCSI for VM storage, not NFS, it is usually faster, though NFS has some administrative benefits in some cases. You could try to configure that on Core and see what happen. If that work reliably, then the problem may be in NFS code. But once we know it is NFS, we would need some sort of problem reproduction, that still may be difficult.

Hello Alejandro, as I mentioned in the previous update, the server in question has been working for more than 15 days without problems, the NFS resource mounted on the ESXi server continues without errors, the disk scrubs are completed without problems, there are no python errors , no unexpected restarts, no excessive memory consumption by pythone like before. The Win10 VM installed on NFS from the ESXi works without problems. Perhaps the problem is due to an incompatibility of some server component in the BSD-based system? If you need any additional information, do not hesitate to request it. Thanks in advance.

Hi Alexander, I understand, don't worry. I hope you can find something. I must mention a test that I have carried out, using the same components in its entirety of the server in question, I installed the scale version based on Linux, the pools were created again and the data was deleted, an NFS share was created again,the resource was mounted on an esxi server, a win10 pro was installed for testing. So far, after almost 4 days of work, the computer has not restarted unexpectedly, there are no python errors, everything works without problems, I will wait a week of use and I will update you. This did not work like that with the bsd-based core version. Thank you

I’m sorry, Leonardo, but your case is difficult. Memory corruptions always are. What we see in debug are consequences of the problem, not the cause, and I don’t have even direction to start searching. I am trying to look for other reports that may be shared, but it does not give any time lines, sorry.