VMware Cloud Community
Torturo
Contributor
Contributor

error "esxi 5.5 quick stats is not up to date" following the installation of ESXi5.5

Hi All,

I am french so find same text in french version at the bottom.

ESXi 5.5 was released 2 days ago.

I have installed it today for a customer wanting it.

We have this error message once the infrastructure was installed, vcenter 5.5 + ESXi 5.5, in the "summary" tab of all ESXi in vsphere client :

"esxi 5.5 quick stats is not up to date"

Because in the middle our migration we had 4.1 and 5.5 ESXi, and both got the message, I assume it is coming from vCenter 5.5.

vCenter virtual appliance 5.5 is being used here.

ESXi are HP DL380 G7 Proliant servers.

Does somebody or VMware has any workaround please?


Regards,

Florian

________________________

Bonjour à tous,

ESXi 5.5 est sorti avant-hier.

Je l'ai installé aujourd'hui pour un client qui désirait cette version.

Nous rencontrons un message d'erreur une fois l'infrastructure installée, vcenter 5.5 + ESXi 5.5, dans l'onglet "résumé" de chaque ESXi dans le client vsphere lourd :

"esxi 5.5 quick stats is not up to date"

A un moment nous avions à la fois des ESXi 5.5 et 4.1 dans notre vcenter 5.5, et ils avaient tous le message d'erreur, donc je suppose que cela vient de vCenter 5.5.

J'ai installé la virtual appliance du vcenter 5.5.

Les serveurs ESXi sont des serveur HP Proliant DL380 G7.

Quelqu'un a-t-il déjà rencontré ce message, et avez-vous une quelconque information sur sa sa cause et sa résolution svp?

cordialement,

Florian

13 Replies
iforbes
Hot Shot
Hot Shot

I'm geting the same error, except I only installed vCenter 5.5 which is managing my ESXi 5.1 hosts. Any comments?

Reply
0 Kudos
ksram
VMware Employee
VMware Employee

Hi Florian,

vpxd logs should record one invocation of task fetchQucikStats per host per minute. If the these stats are not available for over 2 minutes, it is marked as an issue on the host. Can you please check this? By the way how many hosts does this vCenter manage?

Thanks

Ram

Reply
0 Kudos
sdsix
Contributor
Contributor

same problem here from 5.1 to 5.5 via VUM

Reply
0 Kudos
Torturo
Contributor
Contributor

Hi Ram,

our vcenter appliance is managing 3 ESXi 5.5.

I will try to check what you said tomorrow, I was too busy today....

Bye

Bonjour Ram,

notre appliance vcenter gère 3 ESXi 5.5.

J'essaierai de regarder ce que vous m'avez dit demain, j'étais très occupé aujourd'hui.

A bientôt

Reply
0 Kudos
ksram
VMware Employee
VMware Employee

Seems like there is check enabled on vSphere 5.5 relating to qucik stats collection. These quick stats are collected every minute from every host in the inventory. This has not changed from 5.0/5.1, But a check to verify the collection process is enabled by default in 5.5. The verification is done  by checking recent successful collection time. If this time is more 2 times the interval (1 minute * 2) event is generated to say quick stats is not up to date.

Seeing this event can mean any of the following,

  • One sequence of quick stats collection is taking longer than 1 minute. So  some hosts are not been polled for over 2 minutes
    • To validate track the log lines containing "Invoke done (HOSTNAME, vpxapi.VpxaService.fetchQuickStats)". There should be one for every minute for each host. If the difference between two successive invocations is more than 2 minutes then you know the event will be generated.
  • Host is not able to send valid quick stats
    • Check the vpxa logs on the ESXi hosts. (BEGIN/FINISH tasks for vpxapi.VpxaService.fetchQuickStatst)
  • Host is disconnected
    • A host disconnected event should co-inside the not up-to-date event
  • Host is just added to inventory
    • Quick stats Not refreshed yet. So you know
  • Host delays the completion of sending the stats
    • Check the vpxa logs on the ESXi host  (BEGIN/FINISH tasks for vpxapi.VpxaService.fetchQuickStatst)

The quick stats data is mainly used for

  • Real time statistics displayed in the client (Periodic Performance stats collection is not related to quick stats )
  • DRS uses the statistics to measure the load and suggest recommendations

Currently VMware engineering is investigating the issues. At this moment  mitigating or suppressing the event seems to be only solution. I will post once I have some confirmations on this.

Thanks

Ram

ksram
VMware Employee
VMware Employee

I posted the workaround in Re: vCenter 5.5 "Quick stats on <hostname> is not up-to-date"

As I suspected earlier at this moment we can set the vCenter to behave the old way (5.0/5.1), That is suppress the waring. This will mask the config issue and event reporting. It is done by configuration changes to vCnter. Following changes to the configuration file (vpxd.cfg) is required and vCenter need to be restarted to changes to be effective.

Little long post. But Thought required.

There two methods to change the advanced settings of vCenter using either Client or Manual Edit of configuration file.

Using Web Client to update the settings.

See Documentation:vSphere 5.5 Documentation Center 

  • You can use Advanced Settings to add entries to the vpxd.cfg file, but not to edit or delete them.
    • Prerequisites:  Required privilege: Global.Settings

     Procedure

  • Select the Manage tab.
  • Select Advanced Settings.
  • Click Edit.
  • In the Key, type a key: vpxd.quickStats.HostStatsCheck
  • In the Value field, type the value for the specified key.: false
  • Click Add.
  • Click OK.
  • Repeat the same above steps to add the key: vpxd.quickStats.ConfigIssues with the value of false

Changes will take effect after a reboot.

For brave there is a manual edit method of configuration file

Please note that we are changing the vCenter configuration file here. Make sure to make a backup of the file before proceeding.

Navigate to the vpxd.cfg file.

By default this is located in:

  • Windows-based vCenter Server: C:\ProgramData\VMware\VMware VirtualCenter\
  • vCenter Server Appliance: /etc/vmware-vpx/

Procedure

  1. Backup the existing vpxd.cfg file. Do not skip this step.
  2. Open the vpxd.cfg file using a text editor (Windows) or VI editor (VCSA). For more information on using VI editor, see   Editing files on an ESX host using vi or nano (1020302).
  3. Add the following parameters between the <vpxd> tags:

  4.              <quickStats>
                    <HostStatsCheck>false</HostStatsCheck>
                    <ConfigIssues>false</ConfigIssues>
                 </quickStats>
     
  5. Save and close the configuration file.
  6. Restart the vCenter Server service. For more information, see:

A Knowledge base article should be available soon.

Thanks

Ram

mauser_
Enthusiast
Enthusiast

I have tested this (through the fat-client because webcliënt freezes as "loading" while choosing advanced settings".

Rebooted vcenter, and error is again.

This is working well.

This is a good working work-arround. Will there also come a real solution for this issue?

Thx

Reply
0 Kudos
ksram
VMware Employee
VMware Employee

VMware engineering is working on a fix.

Thanks

Ram

Reply
0 Kudos
PaoloKappa
Contributor
Contributor

Just toggle disable/enable HA and it works again...

... not for a long time, but it works Smiley Happy

Paolo Caparrelli GOLINE SA http://www.goline.ch
Reply
0 Kudos
admin
Immortal
Immortal

This looks to be a known issue in vSphere 5.5 and there is a permanent fix which has been documented in the following VMware KB 2061008. Restarting the management service will not prevent the warning message from coming again, please refer to the KB for the solution.

I do not know about you, but I like to have a clean environment and I get annoyed when I see warning/error messages in the UI. From what I can tell, vCenter Server was able to collect the "quick stats" from the ESXi host but perhaps there was a communication problem at some point or just a glitch?

In any case, is is pretty easy to fix the problem, you just need to restart the management service on the ESXi host and this will force a refresh of the stats. You have three ways of doing this:

  1. Restart management service using DCUI connecting to your ESXi console
  2. Restart management service using DCUI via command-line
  3. Restart management service using command-line script

Option 1 is pretty straight forward and both Option 2 and 3 can be performed on the command-line via an SSH session to your ESXi host if you have enabled SSH.


Option 2 - To launch the DCUI, just type "dcui" on the command-line and you will be able to interact with the DCUI as you normally would from the console and restart the management service:

Option 3 - To restart just using the command-line, you can run the following command to restrat the management service:
/sbin/services.sh restart
- See more at: http://www.virtuallyghetto.com/2013/09/how-to-get-rid-of-quicks-stats-on-hostx.html#sthash.uXIRZ4LO....
Nettweets
Contributor
Contributor

I'm using vSphere 5.5 and experienced this issue. 'Reconfiguring for vSphere HA' on the specific host cleared the warning for me.

Reply
0 Kudos
aschwin
Contributor
Contributor

Thanks ! Followed the procedure Fillips described and managed to solve the problem. 🙂

Reply
0 Kudos