System hangs with excessive svc_t values
Storage setup as one large mirrored pool
Storage split into 7 x 1Tb LUNs assigned to M4000
zfs pool on M4000 1 stripe of all LUNS
One zone on M4000 running E busines suite database
Performance is initially excellent, however after 24 to 48 hours uptime system slows / hangs iostat shows v large svc_t values (200-300)
Application becomes unuseable any access to the storage incredibly slow and 7320 gui becomes almost unusable.
Eventually got SCSI transport errors in M4000 messages file.
After bit of investigation on the web.lowered the values of
zfs_vdev_max_pending