Unsolved
This post is more than 5 years old
3 Posts
0
17891
CX-700 and Navisphere - shutting down SAN
Am new to SANs and to the CX-700. Our SAN administrator left the company and I'm left holding the bag. We have a power outage exercise this weekend and I need to know how to shut down the CX-700 in an orderly manner. I understand that I might be able to do this entirely from a console called Navisphere. How do I do this? Help.
Dev Mgr
4 Operator
4 Operator
•
9.3K Posts
0
October 12th, 2010 07:00
The shutdown order is important, but the powerup order is probably even more important, but I'll go over the shutdown order first.
- shut down all servers that use storage on the CX700 array
- if you can, wait ~5 minutes to let the array determine that no more IO is coming in and flush it's write cache to disk (you could also go to the array properties and under the cache tab uncheck the write cache to force flushing sooner)
- power off the SPS (standby power supply) units (this will initiate a shutdown for the CX700 SPE (storage processor enclosure) and once that has powered down the DAE-OS (disk array enclosure-operating system (this DAE contains the OS for the CX700 on part of the first 5 drives)) will power down too
- power off the other DAEs using the powerbuttons on their powersupplies or by powering off the PDU/UPS that they are connected to
- now you can power down your fiber channel switches and any fiber connected tape libraries
To power all of this up:
- power up the fiber switches first and wait ~2 minutes for them to boot up
- power up any fiber connected tape libraries
- power up the DAEs that aren't connected to the SPS using the PDU/UPS power button or their individual powersupply power buttons and wait ~2 minutes after the last DAE was powered up (to make sure all drives have spun up)
- power up the SPE/DAE-OS combo using the SPS powerbuttons and wait ~5 minutes
- optionally you can now check Navisphere (e.g. from your desktop/laptop) to make sure the array is up and running and look under "unowned LUNs" to make sure that no production LUNs are in there.
- power up the servers that use storage on the array
If during shutdown there still was some write cache, the array will have secured that, but if then you have the bad timing that the DAE for which that write cache was intended wasn't online by the time the storage processors came back up, the LUN that that write cache was intended for will be marked as unowned and there will be "dirty cache" messages in the event logs. This is why the power up procedure is more important than the power down procedure. If you do run into unowned data LUNs, and you're running Navisphere release 19 (off my head) or later, you can right click the LUN and select to 'bring online', which is supposed to take the write cache that's secured and commit it to that LUN to allow access to it again.