vsphere data protection services spontaneously stop

[anthonymaw1]

hello gentlemen

i'm running vsphere data protection 6.1 on ESX 5.5 on a Cisco UCS B200 M3 server.

it co-exists with about a dozen other low demand virtual machine workloads.

The vSphere Data Protection OVA appliance was deployed with the default hardware requirements and backs up to a 12 TB SAN volume on a fairly high performance IBM DS3500 SAS SAN.

all the storage performance tests checked out OK.

I'm backing up around 40 VMs overnight in an overnight 12 hour window.

Backups are configured with the default retention periods.

After installation the backups ran great for the first few months.

But then I started seeing more frequent backup failures or sometimes nothing gets backed up.

I noticed that very often the VDP services spontaneously stop and the backups fail or are incomplete i.e. VM snapshots not cleared.

I can restart services both from the console using "dpnctl start all" or from the VDP-configure web console.

But shortly thereafter many services will stop again especially when they are actually started to do something that previously failed.

this especially includes running integrity checks so that i wind up going several months without a complete integrity check.

also i note that the CPU often gets maxed out at 10 to 12 GHz sometimes.

I have not had any luck in the past year trying to get VDP backup to work reliably.

can anyone tell me what I might be doing wrong ?

with thanks from sunny Vancouver Canada....

anthony maw

[daphnissov]

I have not had any luck in the past year trying to get VDP backup to work reliably.

can anyone tell me what I might be doing wrong ?

You're not doing anything wrong. VDP is (was) not a very good product, primarily owing to the fact it's based on Avamar. At this point, though, salvage your losses and pick another data protection product which is both more robust and has a lifespan, because VDP is dead. A community favorite here is Veeam and it seems to be the defacto standard in protection of vSphere and Hyper-V workloads. There are of course others, however.

[vdp4life]

If you had at least some logs or any information we may be able to help. I've ran VDP for years with no problems, but I do know it does have it's limitations on large data sets. I think the most I've been able to backup on VDP was about 8TB on a stable basis. Above that, we had to look into dedicated spindles or DataDomain to handle the backup data.

The most common scenario is always lack of memory on vdp. Or the disks dedicated for VDP aren't fast enough. You can usually see this by monitoring the swap usage on the linux sub system. If the swaps are high, Gsan wont like it and will crash.

Hope this helps.