Multipath Issues on host due to increasing CRC errors on ethernet ports
Applies to
- AFF
- Ontap 9.x
- ISCSI
- RHCOS(RedHat CoreOS)
- Openshift\Trident
Issue
-
Fewer iSCSI paths are observed on the host than expected, leading to I/O errors, failed or pending multipath status, and inability to create pods on worker nodes.
-
The issue is intermittent, with
multipath -lloccasionally displaying “I/O pending” or “failed” across multiple nodes. - Below events are seen on host end.
[Fri Dec 19 19:28:12 2025] sd 3:0:0:1: Power-on or device reset occurred[Fri Dec 19 19:28:12 2025] connection17:0: ping timeout of 5 secs expired, recv timeout 5, last rx 4301139922, last ping 4301145088, now 4301150208[Fri Dec 19 19:28:12 2025] connection17:0: detected conn error (1022)[Fri Dec 19 19:28:12 2025] sd 2:0:0:1: [sdb] tag#67 FAILED Result: hostbyte=DID_TRANSPORT_DISRUPTED driverbyte=DRIVER_OK cmd_age=12s[Fri Dec 19 19:28:12 2025] sd 2:0:0:1: [sdb] tag#66 FAILED Result: hostbyte=DID_TRANSPORT_DISRUPTED driverbyte=DRIVER_OK cmd_age=71s[Fri Dec 19 19:28:49 2025] device-mapper: multipath: 253:1: Failing path 8:64.[Fri Dec 19 19:28:51 2025] device-mapper: multipath: 253:1: Failing path 8:48.
-
LIFs used for iSCSI connections are up and functioning normally.
-
However, the underlying Ethernet ports show CRC errors when running the
ifstatcommand. - These CRCs appear on the receive side.
- Statistics were cleared to rule out historical data, but CRC counts continued to increase after clearing.
