Brennan, if you go to power link and navigate to -

Home > Support > Technical Documentation and Advisories > Hardware/Platforms Documentation > CLARiiON CX4 Series Systems > CLARiiON CX4-240 > General Reference

Select CX4-240 Storage Systems Hardware and Operational Overview

this will allow you to create a customized document which will have power up / down procedures.

Jim

The link that Jim_A gave is a good starting point for getting a actually document for most customer proceedure's.  

Being a retired Marine, i am used to associating proceedures with acronyms.  I dont know where i got this one from but it has worked great and helps me remember.

Shutdown

HATS

H=     Power down all hosts attached to the array to stop I/O   Wait 5 minutes

A=     Power down the array.   Start with the  switches on the SPS.  This will powerdown the Storage Processers and write any data that is in cache to the vault drives.   Once that powers down (around 1 minute), you can unplug the rest of your attached DAE's

T=     Tape Library.   If you have a SAN attached TL, then power it off.  If not, skip this step

S=     Switches.  power them off.

To power it back up,  just follow the steps in reverse order.   Give yourself about 2-3 minutes for S & T.  When you get to the Array step, plug in all the DAE's not powered by the SPS to get them spinning.   THen hit the SPS switches.  Remember that it will take on average about 7-8 minutes for the array to come online and ready to provide host.  You can tell it is ready when the LED's on each SP have stopped flashing.   Then power on your hosts.

Depending on how many hosts you have, your looking at about 5+ minutes + host shutdown time to power down.   About 15+ minutes to power back up.

HATS     remember it:)

Thanks

^ Nice one Kenn.

There are only two things I would add. One, is that for the power up sequence I wait until I can login to the switch before going to the next step, some switches like the Cisco directors take a long time to be fully functional. Two, for the DAEs I wait untill all the disks are online (green solid light) and that there are no failures before powering up the SPS/SPE/DAE-OS.

Those are both good idea's  I normally work with the departmental switchs that dont take as long as the director class switches to come online.

It is usually right when I start stressing out that my Cisco directors are fully booted up.

I would always manually disable write cache and allow it to get pushed down to the disk.

then power down.

Hi Brennan,

Please check this also.

i have use this procedure, is it valid?

1. stop i/o hosts

2. power off fiber channel switch

3. because there is no transaction i/o, i poweroff aditional DAE, not DAE OS.

4. turn off sps switchs, and system goes down, cpu and OS DAE.

to turn on follow 4 to 1 steps. doing step 4 and 3 at the same time. then wait 10 minutes and continue with 2 and 1

is it wrong or valid.?

No worries, I forgot to also mention that you should do the reverse when powering the array back on (i.e. power up all the non-OS DAEs first, let them settle and then turn on the SPS switches to boot the storage processors)

ok, thanks for the info

Hi maynor,

It is preferrable to turn of the SPS switches before shutting down the non-OS DAEs as otherwise the array will suddenly lose communication with all the DAEs (i.e. swap steps 3 and 4)

thanks Scooba, i will do your recomendation.

, sorry is not my post,  and can not asign points.

mc

Hi Brennan,

The advice I gave is the same as the documentation, it's just that it's a bit unclear. When the procedure above states (Step 6):

"Once this is complete and the SPS units have turned both themselves and the power supplies connected to them off, any other chassis can be turned off by using the power supply switches or removing cables from their AC source."

It is trying to say that you should then power down the remain DAEs (after the SPS has switched off, and the storage processors and DAE-OS are powered off).

Basically try to think of it this way, if you unplug the power to all the DAEs before you power down the SPS and DAE-OS, then FLARE is going to think that those DAEs are faulted. If you turn off the SPS (which causes FLARE to shutdown the storage processors and cuts power to the DAE-OS) and then unplug the remaining DAEs, FLARE will not ever report that it has lost communication with the non-OS DAEs.

An easy way to remember the procedure is to walk through the logical layering - power down (or stop IO from) the hosts first [top of the chain], then the switches (if necessary), then the storage processors (via the SPS switches, make sure it shuts down completely), then unplug the remaining DAEs [bottom of the chain]. To power on, reverse this order.

Scooba,

Am I missing something or is your recommendation opposite of the documentation?

"preferrable  to turn of the SPS switches before shutting down the  non-OS DAEs as  otherwise the array will suddenly lose communication with  all the DAEs  (i.e. swap steps 3 and 4).

Ken,

I like your way of remembering things, sometimes it's the simple stuff that saves you in emergencies and the last time I had a power outage that would have helped. Thanks!

Here is what I have so far but the question is SPS then NON OS DAE's or NON-OS DAE's then SPS?

1. Stop all I/O to SAN 
2. Wait 5 minutes after powering off the last host attached to the SAN to allow write cache to finishes writing to the disks.
3. Power off Storage Switches (McData 4400)
4. Power off NON-OS DAE’s by manually unplugging them.
5. Power off each SP  using power switch
6. Power down Power Strip, grey first then black.
7. Unplug all power cords from power strip.

Ahhhh, I'm with ya now. Thanks for clarifying.