DataCore Software Community Forum

RE: Increasing SAN LUN behind SANmelody 2.0.4.5

9/2/2010 8:52:02 AM

If the LUN in question is already part of an NMV pool then expanding it on the backend to increase its size in the pool is not a wise operation as SANMelody is protecting the disk. The better option would be to create an additional LUN and present it to SANMelody. Otherwise what you would need to do it split your mirrors, take the disk out of the pool, increase the size, add back to pool, then setup your mirrors again.

RE: Increasing SAN LUN behind SANmelody 2.0.4.5

9/1/2010 10:56:27 AM

Well I guess no one has anything to say on this topic. FWIW we've now upgraded to SanMelody 3 in the meantime.

RE: Monitoring NMV Pool

8/30/2010 7:30:22 PM

I use Opsview Community Edition (Nagios) to monitor a couple of hundred servers/devices with thousands of service checks including SANMelody.

Combined with NSClient++ installed on Windows I monitor Datacore services, NMV Pool free space, AIM Logged data, AIM buffer free space, check for mirrors in recovery, Free Page Table Entries, AIM and HA Event logs and of course cpu and memory.

In addition to monitoring Opsview stores stats out of the box for a easy trend view of things such as NMV pool free space consumption.

Its a really good solution that i can recommend. They offer a Opsview Virtual Appliance for those not familiar with Linux.

Ben

RE: Monitoring NMV Pool

8/30/2010 7:20:36 AM

Hi Mavrick,

You could create a simple script and schedule it to run daily / weekly and leverage the following to get the info / functionality you need.

1. "Blat" for simple SMTP functionality
2. "TypePerf" command line to pull the info you need out of the performance monitor object.

Or..Use something like "Event Sentry" which can monitor system heath including services, performance counters, log files and more.....

Bernie.

Monitoring NMV Pool

8/30/2010 3:14:33 AM

Hello,

is there a possibility to monitor the nmv pool free space?
I have created a trigger on datacore that sends me an email if nmv pool below 300gb. but can't find a solution to sent me a daily email with the info of nmv-pool free space .

thx for your answer.

RE: HP Servers, Dell Servers with Small Form Factor Drives

8/25/2010 9:58:46 PM

You do have a valid concern about if one of your SDS nodes went down, you're running totally on RAID 0. But also remember this, if one of your SDS nodes went down, you're also running on one SDS node remaining as well. I do understand your concerns though- it did take me a while getting used to the idea of RAID 0- it really messed with my religion at first, but when I understood how to use SMY to recover, then I was ok with it. If you're concerned about a failed board, you may want to keep a board as a spare on hand if your budget allows. For that spare board to be of significant help to you, I'd suggest you make frequent config backups of the SMY environment, i.e ERD disks for each SDS node and keep your SMY license files for each SDS node handy.

I know in this environment you really need to be watching capital spent on technology and ensuring a decent return so I understand your constraints where the SAS is concerned. It would be great if you could test your config before purchasing all of the SATA drives- perhaps if you could put your hands on a few SATA drives, then you can use those to test the performance. Anyway, I do hope you're successful in your setup- do let me know how it progresses and if I can provide you with any further information about our setup that may help you, I'd be happy to do so. Take care.

RE: HP Servers, Dell Servers with Small Form Factor Drives

8/25/2010 6:58:28 PM

Thank you for a thought stimulating response.

RAID 0 holds many of the answers.

No write penalties, better utilisation of disk.

My concern was;
What would happen if 1 SDS node failed (ie the entire node went down for 24 hours while waiting on a motherboard replacement).

We would be running on a single remaining SDS node - with RAID 0;
My concern was a disk failure within this group - but if we have a few pools, then failure of a single disk could be designed to only impact (say) 1/4 of the customers environment.

It's then a game of statistics - the Probability of SDS 0 being out for x minutes and then during that time the probability of a single disk failing on SDS1.

If I dont go down the Datacore 'Network' Create a RAID 1 out of the 2 SDS nodes, I'm afraid I am still going to have to go with SATA Drives - due to cost

Typically I never go with more than 5 or 6 drives in a RAID 5 group.

Thus 24 SATA drives for a single SDS node would be: 4 x (4+1 drives in a raid 5), and 1 x (3+1 in a RAID 5) giving me 5 separate RAID groups.

My preference is always an increased number of pools, and quick rebuild times in the case.

My 'SATA' concept is 2 SDS nodes, with 24 SATA drives each gives 48 SATA drives delivering Reads (where both SDS nodes are shared between the various servers).
So 48 SATA drives should deliver performance approximate to 20 SAS drives for read.

For write - I am somewhat dependent on the Datacore cache- and here the write performance - in RAID 5 would be a concern - and potentially 48 Drives in the RAID config I have suggested would be no better than 10 Drives in total in a RAID0 due to the 4:1 write penalty of the Raid4.

If the funds were there; I'd go SAS all the way - but for the moment, I have to assume SATA 7.2KRPM is going to give me 35-40% of the RANDOM write IOP's that SAS-15K RPM will - and I know in this customers environment that they are comfortably running everything on 16 SAS drives in a RAID 5 config - so hopefully 48 SATA will do the job

Regards

Good info

8/24/2010 12:29:50 PM

Very nice site!

Good info

8/24/2010 12:29:42 PM

Very nice site! cheap cialis http://opeyixa.com/qvoxtxa/4.html

Good info

8/24/2010 12:29:37 PM

Very nice site! cheap cialis