Multipath root devices did not fail over after SAN controller failure on Red Hat Enterprise Linux

Solution Verified - Updated -

Issue

  • The production server lost connection in a multipath environment to storage. After we lost a fiber switch a path was completely lost. It should have continued to be available through the other available path.
  • When we lost path from one of the switches, server stayed up but all filesystems became readonly. That shows the path that we were using was dropped. It did not flip to secondary path for the lun.

Environment

  • Red Hat Enterprise Linux (RHEL), including
    • Red Hat Enterprise Linux 5 (RHEL5)
    • Red Hat Enterprise Linux 6 (RHEL6)
  • Hardware must provide an onboard ROM that logs into remote storage and presents the storage to the host as a SCSI device.

    • This has been observed on HP ProLiant BL460c G7 Blades with devices provided by Virtual HBA (presents as Emulex FCoE adapter):
    $ lspci | grep -i Fibre
    02:00.2 Fibre Channel: Emulex Corporation OneConnect 10Gb FCoE Initiator (be3) (rev 01)
    02:00.3 Fibre Channel: Emulex Corporation OneConnect 10Gb FCoE Initiator (be3) (rev 01)
    
    • This could also occur on Fibre-Channel storage or iscsi storage that is configured via a hardware configuration menu.
  • Device-mapper-multipath is either not in use, not correctly configured, or not currently managing multiple paths per device.

Subscriber exclusive content

A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.

Current Customers and Partners

Log in for full access

Log In