PANIC: page fault (supervisor read data, page not present) on VA 0x20 in process mlogd
Applies to
- ONTAP 9
- Automated Non Disruptive Upgrade (ANDU)
- Panic
Issue
- Upgrading from any release prior to 9.12.1 without fix for bug CONTAP-123980 , can result in:
- On a normal takeover - a node to be taken over panics but takeover can complete, no outage, just a panic during takeover of the rebooting node.
- On a P-level-patch ANDU - a node taken over to apply the patch comes up with the old ONTAP version and ANDU is paused-on-error.
- On a major-version ANDU (e.g. 9.9.1 to 9.10.1), a node taken over to reboot (to apply the update) panics and takeover fails, data is not served anymore until the node that paniced has rebooted.
- When a node initiates shutdown (for example during takeover triggered by ANDU), it panics with:
PANIC: page fault (supervisor read data, page not present) on VA 0x20 cs:rip 0x20:0xffffffff8069859e rflags 0x10046 in process mlogd onrelease 9.10.0P1 (C)
- Data outage during during ONTAP major-version ANDU (e.g. 9.9.1 to 9.10.1):
LIFs cannot be hosted:
Sat Apr 22 2023 10:33:01 GMT [Cluster1-n02: vifmgr: vifmgr.lifbeingremoved:NOTICE]: LIF lif_n01_mgmt (on virtual server 44), IP address 10.250.50.50, is being removed from node Cluster1-n02, port a0l-179.
When the node that paniced is up again, LIFs are hosted and data access is restored:
Sat Apr 22 2023 12:45:57 GMT [Cluster1-n01: vifmgr: vifmgr.lifsuccessfullymoved:NOTICE]: LIF lif_ALE_n01_179_mgmt (on virtual server 44), IP address 10.251.42.97, is now hosted on node Cluster1-n01, port a0l-179.