pjapk
Contributor
Contributor

Unable to create a Protection Group - "Array Pair" not visible

HI all, at the final stages of a proof-of-concept with a pair of Falconstor NSSVA's replicating an environment, most of the SRM side of things is now configured but at the point where I'm trying to create a Protection Group, my Array Pair isn't showing up (just a blank area where it should be).

So far I've:

  • Configured an environment around 2 x Virtual Centers (with one clustered pair of ESXi hosts in each) & 2 x SRM servers
  • Installed SQL/SRM & connected the two "sites"
  • I've configured Resource, Folder & Network Mappings
  • I've configured PlaceHolder stores in both clusters
  • I've configured & added Array Managers
  • My replicated test LUNs are showing up perfectly in "Discovered Array Pairs" for each site (within Array Managers).

What else can I be missing? The only thing that springs to mind is patch-levels of the FalconStor appliances (currently 6.15, Build 6164).

Because we have physical FalconStor devices (which is why I'm using these within the lab) I have access to patches if required but the first few I tried installing errored out with no explanation so I don't want to go breaking too much.

As I said, we have an existing physical environment around 6.15 so upgrading to 7 isn't an option in the timeframe I'm working to.

It's so frustrating to be this close but not able to get any further!

Any pointers would be VERY much appreciated!

Regards,

Paul

0 Kudos
18 Replies
harshahosur81
Contributor
Contributor

When you connected the array pairs, it should show a list of the volumes that are setup for replication. If you dont see any then there are no storage volumes that are set up for replication.

If you can see the volumes in the array pair, but not be able to create protection groups, it might be if you dont have any VMs on the datastores that are being replicated.

A screenshot of the same would also be helpful.

Harsha Hosur. VCDX #135; MCTS; CCA
0 Kudos
depping
Leadership
Leadership

Do you already have VMs up and running in this environment? I suggest placing at least 1 or 2 VMs on replicated storage... Also I tend to restart all SRM/SRA services before I start working with SRM, as more than often in the past I noticed this was required.

0 Kudos
pjapk
Contributor
Contributor

Thanks for the responses both. I got all excited as I suspected I may not have had a VM on the replicated storage. I did but it wasn't powered on so gave that a go but with no change.

I've since added a second VM, followed by a restart of both SRM services (i.e. Protected & Recovery sites) but still no change.

I've just taken a series of screen-grabs (although I've added another VM since the included grab) so hopefully someone might see something I'm missing.

0 Kudos
kastlr
Expert
Expert

Hi,

check the vmware-dr.logs of your SRM Servers, they should contain some information about the problem.

You could search for

DatastoreGroupManager

array-pair

to start the investigation of the problem.

Regards,

Ralf



Hope this helps a bit.
Greetings from Germany. (CET)
0 Kudos
pjapk
Contributor
Contributor

Thanks for the pointer Ralf.

I've taken a look and, as a first-time user of SRM, find conflicting info in the log.

On the one hand it says "No replicated datastores found for array pair 'array-pair-3199'" (not sure where it gets 3199 from) but then further down it says "Continuous replication is enabled on device 4" and "CDR resource size = 10240 MB for device 4".

So, I'm still not sure if there's a genuine error.

Can anyone take a look at the attached copy of the logfile and let me know if something jumps out at them?

Paul

0 Kudos
harshahosur81
Contributor
Contributor

I am not sure if your inventory mappings are correct from Site B to Site A. Can you ensure that all the mappings are done the same way you have implemented the inventory mappings from Site A to Site B ?

Also looking through the logs, the message you mention are about replication setup on storage, is the replication setup dual way or one way. because from the logs its only showing one way replication.

Harsha Hosur. VCDX #135; MCTS; CCA
0 Kudos
pjapk
Contributor
Contributor

You are indeed correct, I didn't have B->A mappings but have now. I restarted the SRM services just in case as well but no difference.

I believe I configured the replication both ways (there are timemarks on the recovery site falconstor) but can't be 100% sure so I'm about to rip the drives out & add them again...

0 Kudos
pjapk
Contributor
Contributor

Still no joy, having completely removed & re-created a new VMFS (right down to deleting from FalconStor appliances & creating afresh).

One thing I'm not sure about is, assuming for the purposes of this exercise I'm "protecting" a single VMFS volume (and it's the only volume that exists other than one place-holder volume per FalconStor).

Currently:

  • Protected Site - Hosts vESXi01 & vESXi02, within vCluster01/vCenter01 are only configured as "clients" on SRM-VMFS-01 presented from FalconStor01
  • PlaceHolder01 exists on FalconStor01 & is visible/presented only to vESXi01/02
  • Recovery Site - Hosts vESXi03 & vESXi04, within vCluster02/vCenter02 are only configured as "clients" of the replica of SRM-VMFS01 on FalconStor02
  • PlaceHolder02 exists on FalconStor02 & is visible/presented only to vESXi03/04
  • Neither pair of hosts are configured as "clients" to the other site's FalconStor (i.e. FalconStor01 knows nothing of vESXi03/04 and FalconStor02 knows nothing of vESXi01/02)
  • Although vESXi03/04 are configured as "clients" of the replicated version of SRM-VMFS-01 sitting on FalconStor02, neither host can actually see that storage currently. I'm assuming that this is as it should be and it's the SRM/Array Manager that "magically" wakes up this volume, allowing the two Recovery Site hosts to see it, once it has failed things over. Is this assumption correct or am I missing something here?
  • There are currently two VMs stored on SRM-VMFS-01 and powered-on, running on vESXi01
  • Resource/Folder/Network Mappings are in place from Protected to Recovery & Recovery to Protected sites.

Further to an earlier query re: having replication in both directions, I now don't believe this is necessary/possible anyway, assuming my scenario above is correct, as only the "owning" host(s) will be making any changes to the volume at any given time. It's only when the array manager "flips" to the recovery site when writes are made there, which will then (presumably) replicate back to the Protected Site once it's back up. At least, this is how things look when you look at replication within the FalconStor managment interface.

Does anything look missing/incorrect? I'm hoping I'm simply missing something fundamental here...


Paul

0 Kudos
harshahosur81
Contributor
Contributor

If there is a DR scenario, on Site B, the replicated volume is mounted as a VMFS datastore to the servers mapped in Inventory Mapping ( the backend zoning needs to be in place for this and lun assinged to the servers ). So the data is now written to the volume which has been replicated from Site A. When site A comes backup, it doesnt switch over automatically. If you have replication setup from Site B - Site A on the volume or the failback option selected in SRM , the data is first copied over for data consistency and then a failback is done.

I am not familiar with Falconstore so not sure how its setup. In any case once the array configuration is done, the array pair should communicate and let you create a protection group with the datastores on the replicated storage.

Harsha Hosur. VCDX #135; MCTS; CCA
0 Kudos
pjapk
Contributor
Contributor

I forgot to point out that I was working to a guide at http://communities.vmware.com/docs/DOC-11410 for installation (it seems to be a slightly earlier verision of the FalconStor appliance but the principles appear the same).

Just as I was about to post this I thought I'd double-check my config against the document again and spotted that I hadn't acctually created the "storage volumes" for the FalconStor VM on a different vSCSI adaptor. It was a long-shot as to whether that was the cause but I thought I might as well correct it anyway (so effectively creating all the storage/replication again from scratch) but still no joy.

Once again I can get to the point where it's happy with the Array Pair but just nothing visible when creating a protection group.

I'll have a hard time convincing managment to spend on SRM if I can't get a basic test working in a Lab Smiley Sad

Paul

0 Kudos
GaryJBlake
VMware Employee
VMware Employee

Paul

What version of SRM are you testing with ?

I've just seen your post and interestingly I have the exact same issue with a test environment I'm building, the only difference is that I'm using the EMC Celerra UBER Appliance to provide storage.  I have used the same configuration before with no issue.

Gary

0 Kudos
kastlr
Expert
Expert

Hi Paul ,

and even for my own curiosity, you aren't using nested ESX Servers for your SRM environment?

Because I got exactly the same problem with virtualized ESXi Servers running on latest code.

Regards from germany,

Ralf



Hope this helps a bit.
Greetings from Germany. (CET)
0 Kudos
harshahosur81
Contributor
Contributor

had a similar issue with a customer. the array pair was replicated but the protection groups could not be created. We had  VMs which were powered on (no os no data ). Still no luck, In the end we just had to wait for the replication to complete before we could do anything

Can you kickoff a manual datastore refresh on all your hosts. Try putting in a couple of machines powered on with data on them.try with something like a vMA which only takes up about 5 GB ...:smileyconfused:

Harsha Hosur. VCDX #135; MCTS; CCA
0 Kudos
GaryJBlake
VMware Employee
VMware Employee

Paul

Hope this information helps with your issue:

As mentioned before I was experiencing the exact same issue as you the only difference being that I'm using VNX SRA. The SRA configuration for VNX includes an optional entry for specifing what IP Address to use for NFS presentation, in order to fix my issues I made sure that this entry matched the IP Address I was using when mounting the NFS datastore. Originally I was also using a DNS name to mount the NFS datastore which I altered to be a IP Address.

So just to clarify I did the following in the ordered shown:

- Reconfigured Array setting by adding an IP Address for NFS presentation

- Unmount the existing NFS

- Remounted the NFS datastores using IP Addresses

I've since found out that all SRA adapters that support NFS currently don't play well with DNS.

0 Kudos
pmfna
Contributor
Contributor

Ii'm experiencing the same problem with nested esx servers, is there any workaround ?

0 Kudos
GaryJBlake
VMware Employee
VMware Employee

Are you using NFS ?

Regards

Gary Blake

0 Kudos
pmfna
Contributor
Contributor

Hi,

I'm using ISCSI.

0 Kudos
pallipattu
Contributor
Contributor


Hi Paul,

Did you add your array pairs  in SRA Database?

You might have installed SRA in both SRM servers. You should register array pairs in SRA database.

Go through link::https://communities.vmware.com/docs/DOC-11410

Hope this helps.

Thanks,

Ram

0 Kudos