1

I two servers with about 100 LUNs presented to each from an IBM V7000 (Storwize). Each LUN has 8 paths - 2 FC ports x 2 FC switches x 2 V7000 heads - so about 800 devices. Devices/paths are managed by multipathd. Whenever one of the FC switches or ports goes down (for maintenance), the server loses half othe paths to all the LUNs, but at that point everything is fine. However, when the link comes back up and the missing 400 paths are suddenly available again, the server locks up (presumably IO related) for 30-90 seconds.

Should I be looking at queue depth or other settings here? Also, should I be looking on the servers, or on the V7000?

UPDATE: OS is CentOS 6.6 2.6.32-504.30.3.el6.x86_64, FC cards are QLogic 8G

mapa3m
  • 31
  • 7
  • 1
    You don't mention an operating system for these servers - if you're using ESXi I think you may have exceeded the max-paths (depending on version) which could cause exactly that kind of issue. Come back to us with more details please. – Chopper3 Nov 09 '15 at 17:52
  • Ah! Sorry, completely forgot to add that. Updated. – mapa3m Nov 09 '15 at 18:14

0 Answers0