topic Re: CA synchronous mode and array shutdown in HPE EVA Storage

CA synchronous mode and array shutdown

Gwendling — Tue, 23 Aug 2011 06:48:27 GMT

Hello,

I have a configuration with 2 eva's and synchronous mode replication. I would like to shutdown a complete array to move it in another datacenter ...(down for 1 day)

what happens to the lun access when i shutdown an eva?

since they are in synchronous mode are the write I/O bloqued ? do i have to shutdown both EVA ?

do i have to move the DR group to asynchronous first?

What is a best practice for this senario?

Thank's in advance

Gérard

Re: CA synchronous mode and array shutdown

Johan Guldmyr — Tue, 23 Aug 2011 10:02:35 GMT

Hi, there is a nice little "failover" button in Command View that you can use on the DR groups - to manually fail over the DR groups to the other EVA.

http://h20000.www2.hp.com/bizsupport/TechSupport/DocumentIndex.jsp?lang=en&cc=us&contentType=SupportManual&prodTypeId=18964&prodSeriesId=471572

There is a bunch of manuals there.

For example the "HP P6000 Continuous Access Implementation Guide (T3680-96401, June 2011)" sounds like a good idea. See page 103 for failover.

Re: CA synchronous mode and array shutdown

Gwendling — Tue, 23 Aug 2011 11:15:00 GMT

thank's for answering Johan,

but do i have to "failover" if the source EVA remains operationnal ?

Re: CA synchronous mode and array shutdown

Johan Guldmyr — Tue, 23 Aug 2011 11:17:40 GMT

Are all your vdisks/DR groups source on one EVA?
And the other EVA only has destination DR groups?

Or are they mixed?

Re: CA synchronous mode and array shutdown

Gwendling — Tue, 23 Aug 2011 11:20:20 GMT

Yes!

one EVA is source and the other only destination. It's the destination EVA that we want to move to a new datacenter.

Re: CA synchronous mode and array shutdown

Brad_Hedley — Wed, 24 Aug 2011 02:08:35 GMT

I have had a destination EVA array shutdown, in both unplanned and planned scenarios.

When the destination array is shutdown all IO transactions will be logged on the source array. You will be able to see the logging in Command View under the Log tab in the DR Group properties. If you have not configured the log sizes when the DR Groups were created it will be set to the default value, this is not always the best configuration.

What I would recommend doing is checking the rate of change on the DR Group (eg How much data are you expecting to be written to the vDisks in the DR Group while it is down). Then configure that amount (maybe +20% as the new log size) as the new log size. If you do exceed the log you will just have to re-normalize the DR Group, this is not a big deal, just a risk to the business for DR and an annoyance to the Storage Administrator.

Also be aware that when the destination array is turned back on there may be extra load on the controllers and links between the DCs while the DR Group Logs flush. I have always turned replication back on after hours to ensure that no critical systems are affected.

Hope this helps

Brad

Re: CA synchronous mode and array shutdown

Gwendling — Wed, 24 Aug 2011 06:10:14 GMT

Thank's Brad,

What i didn't understand is the concept of logging in synchronous mode. If we need the acknoledge of the destination EVA to complete the write IO this one will never complete, no ?

i'm missing something !

Re: CA synchronous mode and array shutdown

Gwendling — Thu, 25 Aug 2011 08:36:25 GMT

OK i missed this "failsafe disable"

thank's to all.

Gerard

Re: CA synchronous mode and array shutdown

Uwe Zessin — Wed, 31 Aug 2011 07:36:02 GMT

But FAILSAFE is disabled by default and is only enabled for very specific business requirements.

Did somebody enabled it, because (s)he thought it was a good idea to 'optimize' the settings a bit?