2017-11-13 06:15 AM
initially the RAM in Filer 2 was found to be faulted and was replaced, this Filer then booted but failed to see any disks or ATA enclosures on bus 0B.
placed both Filers into Maintenance mode
Testing of the access to the stack of three AT enclosures consisted of;
On Filer 2
Confirmed port 0C could see the FC enclosures, 0D remained offline when connected to the AT enclosures,
Transposed cables between 0C and 0D, 0D then saw the FC enclosures and 0C went offline. Returned cables to original ports.
Next, transposed cables in 0D between Filer 1 and Filer 2, the disks became visible to Filer 1 using the cabling from Filer 2 confirming the bus from Filer end cable through all three AT enclosures was working.
A replacement controller was installed in place of Filer 2 and fitted with the CF card, 2 x FC HBAs, quad port NIC and NV RAM card from the original Filer 2.
The fault symptoms remained unchanged, any working FC port could see FC enclosures but AT enclosures were only seen by Filer 1. Both controllers still testing in maintenance mode and from Loader> prompt.
The PCI cards (2 x FC HBA and quad port NIC) were removed and tests re run, no change.
Next the NV RAM (cluster interconnect) card and the RLM module were removed from the new controller (Filer 2),
During POST port 0D still did not come online, this was the absolute minimum config for the controller.
NV RAM card and RLM module re-installed and the AT enclosures were disconnected so that the bottom enclosure was in a standalone state, then the test was re-run, still no AT shelf seen.
The CF card containing the OnTap Boot loader was then swapped for a spare which had a lower version of OnTap (came with the replacement controller) in case the Filer’s CF card was in some way faulted.
The symptoms remained the same, No AT shelf connected to Filer 2 was seen.
All hardware returned to original configuration and Filer 1 booted into takeover mode to allow access to filesystem as needed to get the virtual servers and data Vol’s back online.
The above test showed that the physical ports are all working and that access to the AT shelves via the cabling / I/O modules used in port 0D on Filer 2 are good as they work when moved to port 0D on Filer 1. And a known good port (0C) on Filer 2 which sees the FC enclosures also fails to see the AT enclosures.
The controller firmware is the same, as is the OnTap version.
Anyone seen this or have any idea what to check now?