vDP simply does not work

[SCX]

I have almost had it with this product (or at least this release).

I did a very simple Backup of my VM, and tried to do a restore.  The VM is running Ubuntu (if that makes any difference)?, and its VMDK files are around 40GB total.

So far, the  vmdk files of the "Restore" are 80GB+ and growing, with the restore progress meter at 92% and no signs of stopping.

What use is a backup/restore product if you can't trust it?

Off to log another support request...

[LoneCrowe]

Have to agree so far all sorts of problems.  Install didn't go right the first or second time, when I followed the directions explicitly etc.  Had to run the vdp-configure several times,  reboot server and it would ask to run it again.  Once I got it going it was fast in the initial backups that was for sure,  but now it uses 100Mbit of traffic while basically idle and not doing backups at all which is absurd.  Now I can't get into VDP at all via  very shoddy and slow web client. 

I'm sure some of the features are much improved if I could get to actually use them.

This was "as usual for Vmware" one step forward and 2 back.

[danfowler]

Hi SCX,

I wanted to get a little more info from you, is your original VM that you have backed up and are trying to restore thin provisioned?  If so what is the provisioned size?

[SCX]

Hi, yes, there are 2 disks, both thin provisioned, disk one (holding the OS) is 40GB, and disk two is 200GB

I currently have an open support ticket about this.

[SCX]

I am thinking that the appliance just does not work with thin provisioned disks.

It's not an Operating System related issue - I'm having the same problem with Ubuntu and Windows 2008R2.

[futureauto]

I don't have anything to add that will help (I literally just got my first backups to complete after a month of trying yesterday and haven't done a restore test yet), just an interesting side note about the 92% mark.  One thing I've learned about my own troublesome workings with VDP and vmware support is that the progress indicators don't know how to handle more than one virtual disk in either backups or restores.  The first virtual disk always seems to complete at 92% without additional updates when it starts working on the second disk.

[FM-DK]

I join the choir:

vDP simply does not work !

Of all VMware products released the last 7 years, I think this is the worst. Even  vDR v. 1.0 was easier to get to work :smileyshocked:

I have tried to get the 2 TB appliance to work and have read documentation, release notes, community forum and articles on the 'net. I ran into allmost all the problems mentioned: No special characters en vCenter, Datacenter or cluster name or everything will fail, W2KR8R2 backups does not work unless you tweak advanced settings, a silly thing like that you can not enter @ in notification email address unless you copy paste it in, logfiles is not in Windows-readable .zip-format, registration with vCenter fails with appliance no. 2 and so on

But the worst is: It is impossible to vDP to take backup of several VM's several times in a row without problems. I have not been able to get a backupjob to run consistently at any time. I have a job with 10 VM's defined. Sometimes it will succceed with 2 VM's and fail with 8 (And no explaination given unless you travel through MB's of logfiles), sometimes it will backup 2 or 3 other VM's but never more than 4 successes.

The hardware I am running this is on is allmost state-of-the-art, every bit on the compatability list and with tons of free ressources.

I have now abandoned this piece of crap and implemented a very cheap 3rd party solution which works flawlessly.

Regards

André

[SCX]

André, would you mind sharing with us what 3rd party solution you went for? I need to try something else at this point, but all I'm seeing is software that costs $$$. Veeam is good, but I don't like how you need to install "agents" all over the place, and it's overkill for our needs.

[FM-DK]

Hi again

Is it possible to send a PM in this system ? Else make a temporary emailaddress and publish it here and I will email the details.

As a VMware consultant I am not allowed to advertise in this forum about 3rd party products. But I am allowed to email you personally :smileywink:

Regards

André

[SCX]

Yes, just click on a person's username under their image, then click "Send Private Message" from the menu on the right.

[LoneCrowe]

Send me the message too :smileyhappy:

[freaky2000]

I'm joining the choir singing "VDP sucks BIG time lalalala" too.

Quite frequently many jobs are stalled at 92%. Some days (usually when nothing has been added for days) it finishes all jobs in less than 1.5 hours. My backup window has been increased to 12 hours now. Logging is horrible.

Btw, 7-zip, which is free (7-zip.org) extracts the zips fine. Windows doesn't seem to be able to handle this ZIP variant (yes, ZIP does change over time - Windows frequently doesn't catch up tho'). No messages in vCenter though - why would it...

Not to mention the horrible crap with the disk.enableUUID. But no worries, we figured out how to change all 50 VM's fast ourselves. Why would you publish that in your KB. It's much nicer to leave it to your customers to figure it out, or have them shutdown tons of VM's manually, switching the setting and powering them up again. After all, who doesn't add 80 hours of work to set shit right after an upgrade. Oh, did I say 80? I'm sorry, I'm neglecting the fact it has to be reversed again after the next fuck up, err update.

It's quite obvious management is pushing stuff out to fast to compete with HyperV 3.0. Guess what... this misery just gets us there faster. Nice going.

Btw, the manual mentions that when jobs hang on 92%, it's out of storage. I have 48% free on the 1.0TB VDP appliance (according to the reporting), which is far from 1.0TB btw (More like 1.6TB... 60% increase?? really?...) and it's thick provisioned. Then again, should I really trust the value reporting returns? Since hardly anything of the product is reliable, it's not very likely that part is.

I'd really like to tune down the number of concurrent jobs. Anyone know if that's possible? On the odd occasion things do go right, it finishes a lot of jobs in 0.5-1.5 hours. If I can make it run only 2 or 4 jobs at once I suspect I'll have a lot less issues.

[GECADMIN]

Could anyone who got the PM regarding a cheap 3rd Party solution please pass it on.  I will take any recommendations - Thank You.

[SCX]

I have come to the conclusion that "cheap" and "3rd party solutions" are not compatibile. Fair enough, I don't mind paying for software as long as it works and is well supported. The number of hours I have wasted on vDP times my hourly rate is already well past the cost of buying good software in the first place.

I have been looking at vmProtect 8 from Acronis, which is around $500 per CPU "socket". There are a few minor niggles, but so far, it seems to work well and is a heck of a lot faster that vdp, and uses far less resources. The install was an absolute joy compared to vDP, and It can also be installed as an appliance, which is nice.

You can configure local or network disks for backup storage (none of this pre-configured size nonsense). The user interface is "OK" - it's fairly simplistic, but seems to get the job done. They have a 14 day trial.

I looked at VEEAM, but it seems overkill for our needs.

For the record, I have no affiliation with any backup software company, I'm just trying to find something that works.

[LoneCrowe]

Hah.. I looked at Acronis and Veeam.. $3500 is ridiculous for us a small shop with 3 hosts and 2 cpu's per host.  What the hell are they thinking? There is no justification for this gouging.. same with Vmware as well for small shops like us.

This whole pricing cartel the virtualization world has gone to is disgusting.  Makes me want to wash my hands of it and just go back to buying small $400 rack mount servers that did the same job and rsync.

[sm00ter]

Just curiously....

Do you have VM's nested under vAPP/s?

I am having quite the same issues with constant failing backups:

VDP: Backup Job .... failed to backup client xxxxx.  Execution Error:  E30927:An attempt was made to backup a virtual machine client which no longer exists in vSphere...

Or inconsistant successful backups!

I have been working with VMware support for about a month now, and they keep pushing towards a permissions issue, so I changed the vDP permissions to utilize my personal account (as I can do anything in vSphere), and I still had the issues.

Just for fun though, I started removing layers of complexity in the cluster to see where vDP gains the ability to "work".  First I removed the VM's from the vAPP, and to just the base of the cluster, and the backups were successful!!

Then I put them back into a Resource Pool, and tried the backup...successful!!

Once they are added to the vAPP... NO DICE!

Now, this is not a "solution" but it does help in the troubleshooting process, hopefully they can figure this issue out quicker, given the fact that the "issue" has been narrowed down.

Check your layers, and see if minimizing them results in successful/consistant vDP backups!!

Just an idea...

Thanks,

sm00ter

[Suresh Babu Nekkalapudi]

I noticed these issues while using thick client for moving VMs in & out of vApps and Resource Pools

No such issues observed while using web client.

[tman24]

We have multiple clusters here, one running ESXi5.0U1, which I'm still protecting with vDR, the other is running ESXi5.1 (vCenter), but the hosts are still 5.0U1. I've had the odd problem with vDR, but on the whole it works, and one big bonus is that you can use standard CIFS or NFS shares for backups. I was really hoping vDP was going to be great, but we've tried it here, and other than totally flattening the entire cluster during the backup, it simply ate resources and never actually completed, so we killed it off and shutdown the VM (haven't powered it on since). What I really don't like about vDP is that you're forced to used VMDK storage for your backup appliance, and allocating TB's of this is not always possible.

I'm now annoyed that I upgraded vCenter on one of the clusters to 5.1, because vDR will no longer connect to this, so if you want a 'free' solution, you have to use vDP, and seriously, that sounds like a world of pain.

We're now looking at alternatives, and Veeam is meant to be one of the best. Not cheap, but we're going to see what we can do on pricing.

[Suresh Babu Nekkalapudi]

Hi,

When you say alloting TBs of storage for VDP, you do not have to commit it upfront. Deploy a 2TB appliance with thin-provisioning. In that way it uses space on demand.

[tman24]

Yes, but you still need to (eventually) allocate TB's of storage from a VMFS volume directly attached to the cluster, be in NAS/SAN or whatever. Before, I could just use free space from a CIFS share from anywhere on our network.

Why they removed the option to use external storage mapped directly into the VM I don't know. It certainly limits flexibility.

[LoneCrowe]

Veeam seems pretty simple from the free version I've tried.  After looking at some demos from Acronis it looks to have more features and just as simple an interface.  Its slightly cheaper than Veeam too.  Doesn't matter though they all intentionally price their products  similarly on purpose like a cartel.  Its disgusting actually.

[sm00ter]

Update on where I am with the vDP struggle:

The vAPP idea seemed to be earth shattering....until the 2nd day.  Then we were back to some backups working, some not, and at the same time it got a LOT more confusing because there were VM's that successfully backed up (in the SAME Cluster, vAPP, Resource Pool) and some that did not.  In the SAME backup JOB!  Not cool!

Additionally, there was some buzz about it being an authentication issue, but I changed users that were utilized (again), and jobs still fail.

All things considered, I simply cannot recommend vDP as a backup solution to my customers, and those that are thinking about it, I recommend caution before they start "relying" on it to be their backup solution.

Things vDP needs BEFORE it can be a production ready solution:

Better logging

- During the backup

- After the backup

Better disk/storage utilization

- Dedup is great, but let me use any disk location, not a VMDK

- What is up with the storage limits?

Better Web/vCenter integration

- Should be able to see what the status of each VM backup us at a glance, not just that it is protected

- Show an error/notification on a VM if it is in a vDP backup job, and it's backup failed/isn't complete

Better job scheduling capabilities

- Giving me an over all window for backups is not useful, I want certian things to backup before others

- Ability to launch a backup from the VM (Backup-Now!)

This is just a SMALL list of things in my opinion.

[BillysGruff]

The required space for a 2TB backup datastore is ridiculous.  Let me manage my space. 

[gravesg]

Wasted to much time on this product as well. But simply don't have the budget for just about 70% of my shop. The other 30% tier 1 apps comfortably goes into Avamar but cant justify the footprint for image backups for the whole environment. VDP to the rescue, right..??WRONG.

As with all the above issues, I've had several unrecoverable appliance crashes now that VMware support cant even figure out and it becomes a festering escalation ticket to EMC....

[devzero]

for me, this also was a horrible experience, worst vmware product ever.

i put it to trash and replaced it with veeam.

works like a charm and is worth EVERY SINGLE PENNY!