Its now out. Applied it this afternoon. A couple notes:
When it comes back online, it will attempt to upgrade the vpx agents on all your hosts. I found that several upgrade attempts failed, a couple recovered. One host did not, remaining in a disconnected state.
Had to apply the vpxagent upgrade manually for that host's VC connectivity to be restored:
\- copy Program Files\VMware\VMware VirtualCenter 2.0\upgrade\vpx-upgrade-esx-6-linux-40644 to your host
\- run 'sh vpx-upgrade-esx-6-linux-40644' at the console, followed by 'service mgmt-vmware restart'
\- in VC, right-click host, Connect
Another note:
HA gets reconfigured on all the hosts. Same with DRS. We found that DRS started moving things all around (we have it automated), so the upgrade process will potentially generate significant load on your hosts.
HA reconfiguration failed on a few hosts. Removing HA from the cluster, then re-enabling it restored HA.
You may want to consider disabling HA & DRS on your clusters before upgrading Virtual Center.[/b]
Regards,
J
Message was edited by:
hicksj
Hicks, Even with Patch 1 nearly every VC i upgraded i had at leat one ESX show as disconnected.. Then had to deal with the vpxa and service mgmt-vmware restart to get it back..
just did an upgrade and it went fine, had to disable/enable HA though...
I didn't have to stop/start services manually - just disable/enble from VC was enough.
Dave
How about the "Faster (or Otherwise Improved) Client Performance." notice any improvement (please say yes)
uummmmmm, really couldn't say. Seems better, but I am probably looking for it.
Dave
We got a build of patch 2 recently as a pre-release due to the heavy performance problems when we went from VC 1.x to 2.0.1.. We have seen slightl performance gains, but our repository is fairly large. We were getting hammered with issues due to the issues with the stat rollups, which is what this patch is mostly to fix. I'd expect that you'll notice that your performance graphs are \*a lot* faster now
There are still definitely significant strides needing to be made in handling performance, and I make sure and remind my TAM and the folks I'm working with of that every time we talk about it.
Just finished the upgrade here. I took your recommendation of disabling HA and DRS before the upgrade. I think that really helped alot. I was kind of worried about what it would do with all those features turned on.
As far as disconnected hosts, I would say about 60% of them show up disconnected, so I've got some work to do.
We're really hoping the performance is better, but I guess that will remain to be seen tomorrow.
Just completed the upgrade, 75 out of 80 hosts now need to be reconnected.
oh well at least there are all the bugs fixed ive been complaining about since october.
John
75 out of 80 hosts now need to be reconnected.
Ouch! This type of update should not be such a huge administrative burden.
As for performance post-patch, I definitely see improvements
\- Listing out all our VM's per host/cluster/datacenter is very snappy
\- Performance graphs now actually work first try (yea!)
\- even accessing VM consoles seems more responsive
Hopefully it remains this way.
Actually I was mistaken it was all 80!, im still trying to get them all reconected now, its taking ages, hope its not this bad next time, ive another 275 esx 2 hosts im upgrading to 3 over the coming months.
John,
When I saw your post, I thought: "275 more hosts? Holy #@^$"
Then I remember during one of our Buffalo VMUG meetings discussing future large scale rollouts with a gentlemen who stated their UK offices were quite heavily into VMware already. He was of course from HSBC.
Nice environment
- Performance graphs now actually work first try (yea!)
Does this mean that all performance graphs now shows up as they should ? I mean does this solve the issue of those perf grpahs being flatted out at 0 even in daily weekly monthly views ?
Massimo.
Only time will tell, but so far so good.
Is it better to disable HA or place the ESX servers in maintenance mode instead?
Definitely NOT maintenance mode!
That would mean every host in your envrionment would be offline. Just disable HA/DRS and then re-enable afterwards.
I wonder if this will have an impact on the hosts reconnecting properly after the update? All the hosts that were not in HA/DRS clusters in my environment reconnected just fine after the VC update.
Nope... All my servers including my blades using local storage and no HA/DRS failed to reconnect. had to restart the agent for it to work.
There was no hardware vendor difference, both IBM and HP servers had the same problem.
John
The docs say that you must complete this task prior to upgrading. My dba's want to know if this privledge is needed just for the upgrade or is it needed permanently.
Pre-Upgrade Task for Oracle Database Server: The schema used to manage the VirtualCenter objects must be able to use Oracleâs DBMS_LOCK built-in package, which means it requires execute privileges on the package. Prior to installing the upgrade on an existing system, you (or your Oracle database administrator (DBA) must log on to the Oracle Database server as the sysdba and grant the privilege, as follows:
sqlplus system/ as sysdba
grant execute on dbms_lock to vpxadmin;
I think this patch will make my DBA happy, although I think I will let you guys test it out a while longer before I upgrade.
Thanks for all the tips everyone!