Solaris Host support recommended settings in SnapMirror active sync (formerly SM-BC) configuration
Applies to
- Solaris 11.3,11.4 with SnapMirror active sync (formerly SM-BC)
- Solaris SPARC and x86_64 Platform
- ONTAP 9.9.1 and later
Issue
IO error/outage may be observed on Solaris hosts after SM-BC Automatic Unplanned Site FailOver (AUFO) workflows is exercised using default Solaris OS settings
Limitations:
- SM-BC ensures business continuity in case of a cluster/site disaster. In case of a disaster scenario, when the site hosting the primary cluster experiences a disaster, the host multipathing software marks all paths through the primary cluster as down and as part of the site failover, I/O resumes via the secondary cluster paths.
- In the event of a disaster, the SCSI failover characteristics (ALUA) on various paths to the same LUN toggle between Active Optimized (AO) and Active Non-Optimized (ANO)
- With the Solaris OS default settings, Solaris Multipathing (MPxIO/scsi_vhci) fails to enumerate the paths when there is a change in SCSI failover characteristics. Ideally it should re-probe all paths again to figure out if the characteristics of previously probed paths have changed. Instead, it results in a hard I/O error.
- Sample messages expected in /var/adm/messages as illustrated below:
Jul 31 xx:xx:xx host-A scsi_vhci: [ID 721035 kern.warning] WARNING: scsi_vhci: failover contradiction: 'f_tpgs'.vs.'f_tpgs': path /pci@400/pci@1/pci@0/pci@0/fibre-channel@0,1/fp@0,0/disk@w200300a098dfd69d,0