Summary:
UEKR6 kernel with particular versions may result in communication disruption, downtime or database freeze when i/o is interrupted from one server.
Symptoms:
- Database processes hanging after communication disruption with another node in the cluster.
- scsi_eh processes and various database processes reported as blocked in /var/log/messages with messages similar to the following:
kernel: INFO: task scsi_eh_6:2498 blocked for more than 122 seconds.
kernel: INFO: task ora_ckpt_orcl2:13745 blocked for more than 122 seconds.
Affected products:
FlashGrid Cluster or Storage Fabric cluster using Oracle Linux with UEKR6 kernel
Affected versions:
Oracle Linux UEKR6 kernels with versions between 5.4.17-2136.301.x.el8uek and 5.4.17-2136.304.x.el8uek
Root cause:
A kernel driver issue may result in iSCSI operations becoming frozen when i/o is interrupted on a remote target server.
Resolution:
- Keep (or downgrade to) version 5.4.17-2136.300.7.el8uek
- Upgrade to version 5.4.17-2136.305.5.3.el8uek or later.