VMs crashing after update to 5.5 Update 3

[JWester]

Hi,
after updating vSphere 5.5 to Update 3 we're encountering several VM crashes during snapshot removement.
VMware confirmed the bug and they're working on a solution. Currently they don't have a public KB entry.
You can also read more here: https://communities.vmware.com/thread/520379
So please wait with an upgrade if you still run 5.5.
Bye
Joern

[Gostev]

Hi, Joern. Thank you very very much for the heads up!
I really appreciate you letting everyone know about this issue.

[Cragdoo]

Ouch sounds a sore one. As someone who was planning to upgrade to 5.5 update 3, id like to thank you for the heads up. Have shared via twitter to raise awareness.

[btmaus]

Thanks for sharing the info Joern, you saved us alot of hassle as we were looking to update as well!

[dellock6]

So basically 6.0 has bugs, 6.0 u1 has, now also 5.5 u3. QA lately has got some vacation down in Palo Alto...

[tinto1970]

thank you! i was about to upgrade to U3, but i had the good idea to search the forum to see if Veeam already supports that.

What an ugly bug!!! There's really something wrong in VMware, today. Ok, probably i'd never had experienced the bug because i always keep VM in "/" folder in the datastore (see posts in the vmware community), but that's really bad.

[Patschi]

Luckily I updated my ESXi about three days ago from 5.5 U2 to U3... However, I was lucky and the bug didn't occured yet on my environment, because the VMs are also saved in the root-folder of the datastore like tinto1970 does it too.

There is a new interesting post in the VMware community thread:

Last night engineering found the root cause. They will need to produce an express patch.
We have asked how long are we expecting to wait for this, I will follow up once a response has been received.

(Source: https://communities.vmware.com/message/2543893#2543893)

[jorgedlcruz]

The VMware KB looks like a joke, but it's not. You need to rollback your ESXi Host - http://kb.vmware.com/selfservice/micros ... Id=2133118 Unless for now, I think it's the first time when I saw something like this from VMware, telling Costumers to rollback the entire ESXi.

[guitarfish]

Veeam backups use snapshots, right? So would running a backup potentially result in this error? I have 6 production hosts, all of which were upgraded to U3 over the last week. Sigh.

[Gostev]

That is correct.

[guitarfish]

Unbelievable. I'm sorry, but heads should roll for this.

[lightsout]

Yeah, this is just incredible. We've got 6 u1 with the NETDEV WATCHDOG issue, and 5.5 U3 with the crashing snapshots issue. Can't believe VMware's QA is so bad. 5.5 U3 also fixes security issues too, so if you back it out, then you've got known security problems.

Although it does appear that having folder on your datastores seem to be a cause, and I don't do that, I guess like most people.

Just upgraded to U3 as well this week too. Not seen the issue so far (touch wood).

[JWester]

We got information from VMware-support that they'll have Express Patch 8 ready on October 15th. 10 days to wait. :-/

[tinto1970]

it would be nice to know if the vCenter will be updated too or not. I guess they will release only a patch for the ESXi but i'd like to be sure before performing the annoying vCenter update

[dellock6]

So far seems the patch will target only ESXi, I don't see why vCenter would be involved. But obviously confirmation will come only when the patch will be available.

[mhetrick]

I upgraded our ESXi and vSphere Center to 5.5 update 3 last week before seeing this, but have not had any trouble so far. We run everything across iSCSI and all virtual machines are stored within the root folder of each iSCSI target folder. We also don't have Enable VMware Tools quiescence turned on either. This thread suggests trying to turn quiescence off and see if that helps. We've never had that turned on, so maybe that's why we haven't experienced any issues. https://communities.vmware.com/thread/5 ... 0&tstart=0

[lightsout]

I was told we'd see a patch this week actually.

I'm still running Veeam with ESXi 5.5 U3, and I have not seen the issue so far thankfully. But I do agree with Gostev's comments on his weekly digest that VMware's QA recently has been awful.

[jsprinkleisg]

Yeah, this is just incredible. We've got 6 u1 with the NETDEV WATCHDOG issue, and 5.5 U3 with the crashing snapshots issue. Can't believe VMware's QA is so bad. 5.5 U3 also fixes security issues too, so if you back it out, then you've got known security problems.

I'm guessing the snapshot issue is caused by something in patch ESXi550-201509201-UG. The OpenSLP Remote Code Execution security issue is fixed in patch ESXi550-201509101-SG. I would assume the security patch can be applied separately, thus avoiding the snapshot issue. Can anyone confirm this?

[Zew]

UPDATE...

This Snapshot bug has been fixed in 5.5u3a

Check it out here!

http://pubs.vmware.com/Release_Notes/en ... notes.html

http://kb.vmware.com/selfservice/micros ... Id=2133824

[lightsout]

Well spotted. I wonder what other issues this patch introduces? ;(

[dellock6]

Be careful what you wish for...

[neil.clarke]

so whos installed it then?

the bug is only affecting one of our servers at the moment so i am holding off as long as i can.

[frinnst]

Installed it late last night on our internal infrastructure and reenabled all our backups.

No issues to report.

[lightsout]

Likewise, I upgraded from U3 last night too, and no issues so far.

[Zew]

I'm here to report, I have updated vCenter 5.5u3 and updated my hosts to ESXi 5.5 u3a.

I didn't make any changes to my jobs, moved one VM into the new host, test a backup job.. worked as usual.

So far so good.

[tinto1970]

ESXi 5.5 U3a released yesterday

[PaulNSW]

So is ESXi 5.5 U3a safe to upgrade to now? We've been avoiding v6 due to the constant critical bugs that are found every week

[Patschi]

ESXi 5.5 U3a should be safe, yes. I didn't heard any issues about yet. However, I'm still running ESXi 5.5 U3 (without the patched snapshot bug) and even this is still running fine in my lab.

And: Still avoiding ESXi 6.0 may be an good idea, because: http://kb.vmware.com/selfservice/micros ... Id=2136854. This should also affect Veeam - as far as I know Veeam is also using CBT (Changed Block Tracking).

[foggy]

Yes, Veeam B&R uses CBT and there's a dedicated thread regarding this issue.

[PaulNSW]

Yeah, I saw that in the forum digest email - Thanks Gostev!

I guess I'll give 5.5 U3a a go then, thanks for the info