VMware Cloud Community
dborgill
Enthusiast
Enthusiast

Upgrading from VC 2.0 and ESX 3.0.2 - Stability?

So we have one old Virtualcenter 2.0 server and about 15 ESX 3.0.2 hosts.. we are debating whether to put VC 2.5 and ESX 3.5 right into production. Have there been any show stopping issues or would people recommend this yet?

We mainly are interested in Storage Vmotion and Update Manager.

Reply
0 Kudos
10 Replies
hicksj
Virtuoso
Virtuoso

Same boat here... similar environment, similar interests in feature updates.

However, given the amount of issues I've seen documented here in the forums (HA spazzing out, DRS loops, VC database conversions gone wrong, etc), I can't rightly justify our moving forward. We'll wait until SP1 at least. I spent 6 months working with VI3 prior to upgrading our VC1 / ESX2.5.x environment, and due to problems I saw then, we chose to remain at ESX2.5 until I could properly test 3.0.1/2.0.1. The new features then were SO darn tempting too... I'm hoping we see updates this quarter.

Reply
0 Kudos
dborgill
Enthusiast
Enthusiast

Good to hear we are not the only ones in that boat Smiley Happy

Really, we have moved very quickly on all VM products in the past but we have SO many production folks using VC and so many critical systems, it's hard to jump the gun now. Any experiences positive or negative would be great.

Reply
0 Kudos
RParker
Immortal
Immortal

Well I am a risk taker. I had ESX 3.5 installed along with VC 2.5 within 24hours of the initial release of 3.5/2.5. We were already at SQL 2005 SP2, so that' helped, but we haven't had any problems.

My biggest problem was 3rd party software, they don't seem to be up to the minute as I hoped, but we are using Beta Versions of the products (backup) and everything seems to be fine now. VM Ware was perfect, I investigated sVmotoin and update manager, and everything seems to be working fine.

We use 3i on one of my machines (so I can thoroughly test everything) and I must say I really like 3i. If you don't have a need for a service console, I recommend it. Of course there are 2 problems with that, bios hardware updates, and command line (such as updating HBA firmware and HBA configurations) that has been a challenge, but luckily we haven't had to rely on this yet.. but there will be a time that we need to access a console, but right now it's great.

I like 3i because of the built in hardware monitoring which in our case obviates the OM software for Dell, which eliminates another layer of programs, which I feel is better.

I have to say I give high marks for 3.5. It's faster, VC 2.5 is smoother (it has a few bugs, but easy enough to overcome) and it's definately better.

If you have a 64-bit workstation/laptop you can't use VIC, that seems to be biggest impetus.

I say go for it.

Reply
0 Kudos
serracon_suppor
Enthusiast
Enthusiast

We moved quiet a lot of esx hosts to 3.5 and vc to 2.5. When they are working, they are working fine, but it is right, getting it to work can be some pain. as mention by the previos poster we also encountered problems with

- HA

- VC Database

- VCB

- Some Kernel issues regarding Hardware accessing

only the mentioned DRS loops have not occured.

So this task is definitly nothing to be done in an afternoons timeframe. always keep fallback options open and do it step by step.

Reply
0 Kudos
dborgill
Enthusiast
Enthusiast

To give more details, we have two large clusters and we use both HA and DRS in the clusters. We have skipped 3i altogether because we use the service console quite a bit and it really doesn't seem to give us much benefit over ESX 3.x

Can you post more details about the issues you have faced?

Reply
0 Kudos
NorbK
Enthusiast
Enthusiast

Ok, I'll have to weigh in here.
We have had numerous issues on a system that has been near flawless prior to upgrading VC and ESX. The biggest issue is if you have any clusters that are heavily used on resources. In these cases, all you need to do is put one of those hosts in maint mode and watch the DRS sprial of death. This is totally unacceptable and identified by VMware who says it won't be fixed until the rollup update coming possibly next month. Now we have other clusters that aren't heavily used in the area of resources that are just fine. But if you utilize a cluster with DRS and make use of that hardware's resources, I definitely would NOT upgrade until the rollup comes out.
Besides the DRS issues, we have had hosts randomly pop up with errors on one of the heavily used clusters. These errors range from DRS errors that magically go away, to vim.fault.notauthenticated errors, to no errors but vmotion failures, etc. Many times restarting the management services makes these go away, sometimes not. Most times maintenance that needs to be done at the local ESX level can be done by directly connecting to the ESX server with the VI client, bypassing VC. Esp when machines can't be shutdown from VC, etc.
The VI client also gets much slower with the 2.5/3.5 release. Be prepared for lots of waiting for this thing to update now.
With that said, the reason we upgraded were for the newer features. They've done some great things with this upgrade, however, the DRS issues have forced us to put our most heavily used cluster into DRS manual mode which has overshadowed the new features. One thing I was upset with is that DMotion (aka storage vmotion) cannot be done from the VI client!!! Command line to do something they practically already do with vmotion. Big let down there. And then let's make it a convoluted command line where people are already starting to write interfaces for it! This whole upgrade seemed about "getting it out the door" to me unfortunately. I hope VMware never does this again because in the long run, its going to leave a mark....that hurts.
dborgill
Enthusiast
Enthusiast

Wow.. thank you all for your input. With almost 200 production VMs and 15 ESX hosts, I think we will put this upgrade off for a bit. I'll still test it in a controlled environment but it seems a bit too buggy still.

Still though, I'd like to keep it alive for a bit to ask a few questions. Did anyone go from a SQL 2000 to SQL 2005 server? I'd love to hear any tips, tricks or issues you may have had.

Reply
0 Kudos
dborgill
Enthusiast
Enthusiast

Can someone tell me if the issues they have been facing with DRS and HA just with the VC 2.5, ESX 3.5 or only in a combination?

Reply
0 Kudos
Milton21
Hot Shot
Hot Shot

only errors I have seen during upgrades are the VC client for the ESX clients might not update... Smiley Sad You will have to reboot. The DB upgrades need planned out and backups taken. Also have a solid rollback plan... But this goes for any major upgrade. I have had 0 problems with HA and DRS.

Reply
0 Kudos
NorbK
Enthusiast
Enthusiast

Can someone tell me if the issues they have been facing with DRS and HA just with the VC 2.5, ESX 3.5 or only in a combination?

Our upgrade was both VC and ESX which were near current versions (I believe 3.0.2 didn't have the last round of patches). We went straight through from VC > ESX > VCB proxy. Now let me stress that I haven't had any issues with DRS or HA on our lightly used clusters. This only affects clusters that would normally move a VM here and there due to resource contraints. If you don't keep your ESX servers tight or if your VMs never work heavy enough to cause a DRS migration, then you probably won't see these issues.


For more detail, see the thread:


http://communities.vmware.com/message/848897

</div>