NF8480M5 / PM8204: RAID5 Device Offlined After a Disk Failure

Source incident

On an NF8480M5 with a PM8204 RAID controller, the source reports a RAID5 virtual disk exposed as /dev/sdb. BMC recorded a DISK7 drive fault/media errors. The OS then logged metadata I/O errors and rejected I/O to the offlined SCSI device 0:1:0:1. The installed smartpqi driver was 0.9.14-100.

smartpqi 0000:17:00.0: resetting scsi 0:1:0:1
sd 0:1:0:1: [sdb] Medium access timeout failure. Offlining disk!
sd 0:1:0:1: rejecting I/O to offline device

The source says hot replacement of the failed drive did not restore OS access, while a later reboot did. That is its incident outcome, not proof that reboot alone repairs the disk, filesystem or driver issue.

Related English vendor evidence

The related Red Hat excerpt describes hardware RAID1, with different PCI/SCSI addresses. It supports the reported driver/reset symptom pattern but is not the same NF8480M5 RAID5 capture. A member-disk failure and loss of OS access to the whole logical device are different failure layers.

The source images list RHBZ1666912 and historical RHEL7 kernel baselines: 3.10.0-1062.el7, 3.10.0-957.27.2.el7, 3.10.0-862.37.1.el7 and 3.10.0-693.50.3.el7, with their original 2019 advisories/dates. They attribute erroneous offlining after reset to communication between smartpqi error handling and kernel retry logic. These archived baselines are not current upgrade recommendations.

Source remedy and scope

The source recommends replacing the disk with media errors and upgrading to a supported matching driver/firmware combination. It names smartpqi-2.1.12-055.tar.bz2 and PM8204 firmware 3.53 as its period-specific package/target. The stored article does not provide a usable package download, and this translation does not invent one.

Check the exact OS kernel, controller and OEM qualification matrix before updating. A driver-only or firmware-only change may not address the applicable kernel interaction. Preserve logs and confirm array/filesystem state and restored application I/O after the supported remedy; replacing a physical disk does not automatically bring an offlined OS device back online.

Official English reference: Red Hat smartpqi device-offline issue. Its public issue/environment text was checked; the complete subscription-gated current remedy must be verified through the applicable support entitlement.