Hey there,
I'm very impressed of the new capabilities VC 2.0 provides. But I have a problem with logging of performance data. It seems that VC is only logging data when my client is running. All the other time the lines of the charts are going to zero. Do I have to configure this at the server?
What do I have to change?
Thanks for your help!
daniel
Yeah, makes sense... I guess...
Weekly and monthly stats should still show something unless the time period the server has been up is minimal. I know on the first ESX 3.0 server I set up that I have a pretty good portion of both and even the one giving me problems has some data there.
The only week/monthly data I see, are for the times during the day when the VC client was connected, all other times the data is there - but flatlined.
Yeah, that is kind of what I get on the one that is messing up. But at least something does show.
Monday morning:
I see the same behaviour as bevor my new installation of VC. What I will do next is a complete fresh install of the whole machine OS included.
I will let you know afterwards but I'm afraid I can't do this not before next week. To much other things to do 😕
I upped my Statistics collection thread limit[/b] to 6 on Friday and all three of my servers have stats through the weekend. I chose 6 based on 3 servers and hoping I wouldn't have to create too many.
I haven't yet checked the individual VMs to see if they have all the stats but it looks like maybe that value needs to be higher the more ESX servers you have being monitored. Maybe 2x the ESX servers?
I upped mine too and now statistics are show for the last three hours. Seems that this looks much better now. I will observe this ...
Weekly and monthly stats should still show something
unless the time period the server has been up is
minimal. I know on the first ESX 3.0 server I set up
that I have a pretty good portion of both and even
the one giving me problems has some data there.
I agree that this should be the behavior. It was that way in VC 1.x. If I started a new VM and took a look at the performance charts a couple days later, both the weekly and monthly would show the couple days the VM had been active.
However, this morning I still can not see the weekly charts. I can do a custom chart for 5 days (roughly the amount of time I have had my collection threads set to 16). But anything more than that results in a flatline for the entire chart. I get the feeling that my upgraded database has data in there that VC2 does not know what to do with, which in turn messes up the performance graphs that show data that was collected with VC1.x
To be honest, I have no desire to throw away the last year and half of performance data to see what a fresh database will give me, so I guess I'll just wait until the next release of VC2 to see what happens.
Message was edited by: nwallaceml
\-- added some clairification
Monday morning:
I see the same behaviour as bevor my new installation
of VC. What I will do next is a complete fresh
install of the whole machine OS included.
I will let you know afterwards but I'm afraid I can't
do this not before next week. To much other things to
do 😕
pfuhli,
I do not think a completely fresh install of the OS will get you anywhere. VC is an application that does not integrate with the OS. Also, as was mentioned earlier in this thread, there is a known bug in VC2 that is causing this problem. Based on the fact that both new installs and upgraded installs with a MS SQL database both exhibit this behavior, I suspect the problem is with a stored procedure or with one of the setup scripts that fails under certain database or ODBC configurations. Or, since increasing the number of threads seems to help, maybe there is just a timing or contention issue with one of these components.
Besides, didn't you say you started with fresh installs of the OS and everything else? Just trying to help with an excuse not to do it... I know how things can get crazy when you are really busy.
Nathan,
thank for your advice - I really appreciate it not only for giving me an apology !
There are some other issues with this server which led me to this idea. I'm not quite sure if I really will perform a fresh install. We use the VC server as a central management server also for our physical machines. So there is an instance of Fujitsu Siemens ServerView installed which is mainly collecting statistics and doing monitoring tasks via snmp. We see a very strange behaviour on this machine which we were not able to reproduce at this time. From time to time it comes that this machine is completely freezing. Only a hard reset will wake it up again. The syslogs and the monitoring say nothing. To exclude software issues we think about a fresh install. But it seems that this is an hardware issue.
Following your advice I will wait until the first bugfix release comes out - which will be acc. to VMware support next month.
This mornig I saw some more statistics. At least there is performance data from the host. But when I dig into the special graphs per vm sometimes there not even flatlines 😕
Research in this is no fun when I think of the money we paid for the licenses. Hopefully this will be fixed in the next release.
Does somebody know if there is a list of known bugs - something like bugzilla?
Does somebody know if there is a list of known bugs - something like bugzilla?[/i]
That would be a nice feature to have available. Might divert some questions on the forums too or be a handy way to at least answer the questions. I guess we could always create a topic that does that if one doesn't exist :D. Just link to topics where it is mentioned that it is fixed in the next version.
yes - good idea.
Ive personally had the following (overall):
"Not enough licences" to power on a dual-proc vm - (power on a single proc vm first) - known bug
Loss of historical stats on VC upgrade (tough!) - known bug i think.
Stats collection not starting - (clear the tables down) - known bug i think.
No collection of "CPU/Usage Average Rate" from VM's on esx 2.x - (no fix) - known bug.
And this, No data when client is off-line (up collection stats, poss workaround) - known bug
There is also the 'resources used by templates and powered off VMs' -- fixed in 2.0.1 according to one post. :smileygrin:
I finally looked at each VM on my problem ESX server and it shows minor dropouts on a few of them. The dropouts are minor in duration, maybe 5-10 minutes at the most. The strange thing is, on any server that I had already set to look at the past day on CPU utilization, in previous days, still had that setting and didn't have any dropouts.
yes - i'm seeing short dropouts also. So not a 100% solution, but an improvement.
I have 36 hosts in My Virtual Center and I only get data when I am directly attached to the host looking at performance. Otherwise it shows flatlines or nothing. I have done both the SQL Query and the Statistics collection thread limit is set to 16. Anyone else have the issue where nothng suggested here seems to help?
I also see this issue. Didn't correlate with running the client until I saw this thread.
Any dates on 2.0.1 other than ~September?
Was there any confirmation from VMWARE that these support requests resulted in a BUG ID to be fixed in 2.01 release?