4

I have an MD3000i Dell SAN with dual controllers. Long story short - it managed to muck up some of its own readings on drives and Dell's support told me I could try to shake it loose with a firmware and NVSRAM upgrade which I need anyway. Since I have a dual controller this can apparently be done live and the upgrade process handles shuffling LUNs between controllers as it updates.

Does anyone have experience with this? The Dell technician claimed it could be done and the documentation suggests as much for dual-controller configurations, but it makes me very nervous to do it.

My main questions are:

  1. Has anyone done this successfully or know someone who has?
  2. Are there any pitfalls to be aware of?
janos erdelyi
  • 225
  • 1
  • 4
  • Regardless, this should be done during a maintenance period... – xeon Mar 08 '12 at 00:30
  • Umm, What exactly is the problem? Please consider revising and including error messages and what the suggested fix from the dell tech is. – G Koe Mar 08 '12 at 00:31
  • G Koe, the logs show that slot 14 in the SAN has 3 drives crammed into it, which is not the case, nor is it possible. i went over the support bundle with the dell tech and he showed me where it is showing that. he's suggesting that since need to do a firmware upgrade, that it may kick that loose and fix the issue. if it does not... then power cycle - and THAT i am not currently prepared for. – janos erdelyi Mar 08 '12 at 14:11

2 Answers2

2

I have successfully performed an update like this on thea same hardwareway for very similar reasons on a couple occasions. My experience was good and required no down time, although it was restricted to a couple esx 3 hosts. Definitely need to have good backups and execute during a maitainance window.

Just last night I had a bad firmware update for my HP san and it took three hours to get it cleared up. That included over an hour of downtime for all VMs using this storage.

Tim Brigham
  • 15,465
  • 7
  • 72
  • 113
  • 1
    Agreed. I have done an on-line MD3000 (SAS, not i) upgrade with dual controllers with no issues. The only odd part was that Dell's technicians were more confused on the process than I was. Just read the release/upgrade notes and follow it to a T, and you should be fine. I would definitely recommend doing this off-hours or during a maintenance period as it is not without risks. – Doug Luxem Mar 08 '12 at 03:49
  • sounds good. i do not have as much backup as i would like, but i will definitely schedule this during a low i/o time – janos erdelyi Mar 08 '12 at 14:08
0

I have tried replacing san controller 2 times, 1 good 1 bad.

The first time was with a vsphere cluster, the second time with an oracle failsafe windows cluster.

Both maintenances involved replacing controller, and syncing controller firmware (upgrading firmware on the old controller with the one on the new).

The first time, everything went very smoothly.

The second time, although during the upgrade process no san path was lost, all luns were visible all the time. somehow oracle failsafe was triggered, and primary server tried to transfer role to secondary server but failed, secondary server failed to detected this and pick up the database, resulted a service suspension.

The cause of the second case was still unclear, although misconfiguration on cluster software was one of the possibilities. I would definitely recommend a maintenance windows, just for the sake of stability.

A Lam
  • 1
  • thank you for the feedback. i'm using multipathd on linux hosts to handle the pathing failovers, but that doesn't mean i'm immune to this scenario! – janos erdelyi Mar 08 '12 at 14:09