[Symptom Description]
After the LUN scanning command is executed on the host, the multipathing software of the Linux operating system (such as Red Hat) is used to scan the mapped LUNs. It is found that only some LUNs can be identified, but all disks can be viewed by running the lsscsi and df –l commands.
LUN scanning command: echo "- - -" > /sys/class/scsi_host/${tmpHost}/scan
[Alarm Information]
No alarm is generated.
[Cause Description]
MultiPath cannot delete unmapped LUNs. During the scanning, it delivers read commands to the deleted LUNs. After the storage system returns a message indicating that no mapping exists, MultiPath marks the path as invalid. As a result, subsequent normal LUNs cannot be identified.
[Troubleshooting Roadmap]
1. Check whether the host uses multipathing.
2. Run the df –l or lsscsi command to view all disks.
3. In storage logs, the keyword scsiReportLunsExecute indicates that all LUNs have been reported, for example, [Report 69 LUN to host 0x10000090fae65d57,actual 69 LUN.][SCSI][scsiReportLunsExecute,458][CSD_5].
[Solution]
Method 1: Clear the unmapped LUNs on the host. For details, see section http://www.linuxquestions.org/questions/linux-distributions-5/finding-luns-928777/.
1. deleted mpath device with multipath -f mpath<number>
2. cd into /sys/class/scsi_device
3. then echo 1 > device/delete for each device (based on scsi id)
example:
multipath -f mpath2
echo 1 > /sys/class/scsi_device/5\:0\:0\:2/device/delete
echo 1 > /sys/class/scsi_device/6\:0\:0\:2/device/delete
echo 1 > /sys/class/scsi_device/5\:0\:1\:2/device/delete
echo 1 > /sys/class/scsi_device/6\:0\:1\:2/device/delete
Method 2: If there are a large number of unmapped LUNs, manual clearing is risky. Therefore, it is safer to stop services and restart the host.
[Post-Recovery Check]
All disks can be discovered after multiPaht rescans disks.
[Applicability]
All OceanStor V3 series V3 series versions
