Hyper-V Host Agent Problems
Hi, I upgraded our VMM 2008 R2 SP1 environment to VMM 2012 SP1 RU2 last month which includes a couple of clusters and several standalone hosts.
Since then the VMM agent on the hosts is crashing regularly, every 1-3 hours for some hosts.
I've attached a log from event viewer showing the vmmagent.exe crashing below.
The Hyper-V hosts are all WS2008 R2 SP1.
VMM Server is WS2012
VMM is 2012 SP1 RU2
I've looked at the Hyper-V hosts and VMM server app and system event logs and there is nothing in there of significance.
One thing that seems consistent is the more load the Hyper-V server is under the most frequently the VMMAGENT will crash.
To help alleviate this issue I have set the vmmagent service to always restart after crashing although this doesn;t help when trying to P2V etc as the job fails when the host agent crashes.
http://technet.microsoft.com/en-us/library/gg682077.aspx#BKMK_SupConfigVirtualization
Configuration Manager supports installing the client and site system roles on supported operating systems that run as a virtual machine in the following virtualization environments.
This support exists even when the virtual machine host (virtualization environment) is not supported as a client or site server. For example, if you use Microsoft Hyper-V Server 2012 to host a virtual machine that runs Windows Server 2012, you can install
the client or site system roles on the virtual machine (Windows Server 2012), but not on the host, (Microsoft Hyper-V Server 2012).
Thanks & Regards, Kedar
Similar Messages
-
I have several Windows Server 2012 R2 hosts. The OS is freshly installed, no roles/3d party software besides Hyper-V role). I also have a freshly installed VMM 2012 R2 server. They all are in the same AD domain. I want to manage those Hyper-V hosts
with this VMM instance, but I have troubles connecting those hosts to VMM.
Whenever I try to add a Hyper-V host in the VMM console, the process is stopped almost immediately after it starts. The failing step is "1.2 Install Virtual Machine Manager Agent". The final error message is as follows:
Error (410)
Agent installation failed on XXX.
Incorrect function (0x80070001)
Recommended Action
Try the operation again. If the problem persists, install the agent locally and then add the managed computer.
If I try to install the agent manually copied from the VMM host (C:\Program Files\Microsoft System Center 2012 R2\Virtual Machine Manager\agents\amd64\3.2.7634.0
as of now) by just clicking on its MSI file in Windows Explorer (and agreeing to the Windows prompt to run it with elevated privileges), the installation fails as well with "Installation success or error status: 1603." in the
server Application log.
The only way I found to install the agent is to run the installation from command prompt with elevated privileges. In this case the installation finishes successfully, and I can add the host to VMM by marking the "Reassociate the host" checkbox
in the VMM wizard.
Why is it so? Isn't the agent supposed to install without manual intervention? As I recall, I had no such problem with previous versions of Windows Server and VMM.
Evgeniy Lotosh
MSCE: Server infractructire, MCSE: MessagingI brought up the share because it is off by default with Hyper-V Server as well as a Core installation. Regardless of the Firewall policy involved.
Maybe, but I don't work with pure Hyper-V and Core servers. I have a full-fledged Windows Server 2012 R2 Datacenter.
You're correct that File Server role is not enabled by default, this is my mistake. I open SMB-IN port manually (with the help of a group policy), and Windows considers it to be am equivalent of the enabled File Server role. Nonetheless, ADMIN$ share is
always accessible in my environment (and without it, for example, I wouldn't be able to remotely install things like SCOM Agent which installs without a glitch). Just in case I manually installed the File Server role, and it didn't help.
What you say about multihomed servers is interesting. The servers in question do have multiple network interfaces, but at the time I tried to install the VMM Agent only one of them (a designated host management / network access interface) had a real IP address.
Two other interfaces were non-configured (I apply a virtual switch configuration after the VMM Agent is installed). Just in case I disabled them when tried to install the VMM Agent remotely on the last host, and the installation still failed immediately
after start. Anyways, after the VMM Agent is installed manually the host can be normally re-associated with VMM, so it's doubtful that I have a network issue.
And, of course, there is no 3d-party software on the hosts (no antiviruses/firewalls in particular) except the EMC Unisphere Host Agent (a piece of software necessary for connecting a host to an EMC storage system) and networking drivers. Actually,
this is a pure Microsoft environment.
So I still believe that is has something to do with the Windows Installer inability to configure Windows Firewall policy after being invoked from Windows Explorer with a double-click on a MSI file. But I don't know how to strictly confirm it.
Evgeniy Lotosh
MSCE: Server infractructire, MCSE: Messaging -
Command for silent installation of SCVMM Agent or adding Hyper-v Host to SCVMM
Hello..
I have to add my Hyper-V host using command to SCVMM without access to SCVMM Console.
Plz provide me commands for that.Hi Richa,
I tested in my lab to add a domain joined hyper-v host with following command:
Add-SCVMHost -ComputerName serverlab-01 -VMMServer scvmm2012r2-01 -Credential (Get-SCRunAsAccount -Name "domainadmin" )
The Runasaccount "domainadmin" is create via command "new-scrunasaccount" (during this command it will prompt to input the username and password in my case I added a domain admin account ):
$credential = Get-Credential
New-SCRunAsAccount -Credential $credential -Name "domainadmin"
Actually (in most cases), you can use "view script" in the GUI or at the end of the wizard , this may help us to know which command was used :
Best Regards,
Elton Ji
Please remember to mark the replies as answers if they help and unmark them if they provide no help. If you have feedback for TechNet Subscriber Support, contact [email protected] . -
Host Agent Service critical - Paging file is too small for this operation to complete
Hi all,
at the moment I have the problem that some of my hosts get in the errorstate critical in VMM 2012 SP1.
The error details in VMM are:
Error (2912)
An internal error has occurred trying to contact the hostname server: : .
WinRM: URL: [http://hostname:5985], Verb: [ENUMERATE], Resource: [http://schemas.microsoft.com/wbem/wsman/1/wmi/root/cimv2/Win32_Service], Filter: [select * from Win32_Service where Name="scvmmagent"]
The paging file is too small for this operation to complete (0x800705AF)
Recommended Action
Check that WS-Management service is installed and running on server hostname. For more information use the command "winrm helpmsg hresult". If hostname is a host/library/update server or a PXE server role then ensure that VMM agent is installed
and running.
On the host i cannot connect to FailoverCluster or Hyper.V Manager. After a restart of the host the Problem is gone and the System is working without any Problems for some days.
The paging file is set to 4GB and the OS is Server 2012.
The hosts have 192GB of RAM.
The configuration of the pagfile didn't change from Windows 2k8R2 to 2012.
I didn't find any Errors in the eventlog. Do you have any idea where to check for that?
Thanks,
PatrickThe hotfix didn't solve the Problem for me either. The WMI crash came again after a couple of weeks.
An additional problem with VDS was detected and the following hotfix was installed:
http://support.microsoft.com/kb/2884597/en-us
Also a problem with Clusterservice was found, sadly MS is tagging this handle leak as "won't fix".
Nevertheless one WMI Instance is still growing...
We are testing a hotfix that so far looks promising. Its a private hotfix only right now. But the file is listed as KB494704 you may want to ask your support engineer to look into it. -
Cannot ad Hyper-v host from unstruted domain
Did anyone face a problem like me? When I try to add a untrusted Hyper-V host in to SCVMM 2012 on VMM console, console suddenly disconnect to server. After I reopen up the console, find out process failed at step installing agent.
with error message like this:
1700
The Virtual Machine Manager service on the %ServerName; server stopped while this job was running. This may have been caused by a system restart.
To restart this job, navigate to the Jobs view and select the job in the results pane. Then, in the Actions pane, click Restart.
Command Line Interface: Restart the job by running the following command:\n\nPS> Restart-Job -Job (Get-VMMServer %ServerName; | Get-Job | where { $_.ID -eq "%TaskID;"})
tTry to run this job again and problem still there. gonna be crazy with this!Might be better to post this in the SCVMM forum. Here is one that deals with clustering -
http://social.technet.microsoft.com/Forums/en-US/home?forum=virtualmachinemgrclustering&filter=alltypes&sort=lastpostdesc
.:|:.:|:. tim -
Does Deep Security able to manage Hyper-v through Agent and Agentless?
Does Deep Security able to manage Hyper-v through Agent and Agentless?
Hello,
I have installed VMM 2012 R2, this is also not allowing me to add and manage Server 2012 R2 hosts. However it is perfectly able to manage Server 2008 host.
It gives following error when tried to add Server 2012 R2 hosts:
Error (2916)
VMM is unable to complete the request. The connection to the agent windows2012.test.com was lost.
WinRM: URL: [http://windows2012.test.com:5985], Verb: [INVOKE], Method: [Associate], Resource: [http://schemas.microsoft.com/wbem/wsman/1/wmi/root/scvmm/AgentManagement]
Unknown error (0x80338012)
Recommended Action
Ensure that the WS-Management service and the agent are installed and running and that a firewall is not blocking HTTPS traffic.
This problem can also be caused by WMI service crash. Ensure that KB 982293 (http://support.microsoft.com/kb/982293) is installed on the machine if it is running Windows Server 2008 R2.
If the error persists, reboot windows2012.test.com and then try the operation again.
Any suggestions please? -
Unisphere Host Agent cannot install on Windows 2012 R2 Core
Cannot install EMC Unisphere Host Agent on Windows Server 2012 R2 Core server.
When trying to install into Hyper-V 2012 R2 I get an error informing of the requirement for Feb 2013 updates to Windows 8 and Windows server 2012.Cannot install EMC Unisphere Host Agent on Windows Server 2012 R2 Core server.
When trying to install into Hyper-V 2012 R2 I get an error informing of the requirement for Feb 2013
updates to Windows 8 and Windows server 2012. -
Enabling VM Guest NLB w/Multicast IGMP on 2012 Hyper-V host w/ converged SCVMM fabric switch
What a mouthful.
As short as possible:
WHAT I'M ATTEMPTING:
I'm trying to build a new NLB cluster for a 2008 R2 SP1 Remote Desktop Services farm. And I'm trying to do it the right way, with multicast igmp, not unicast.
The two guest VMs with NLB install converge fine. VIP gets this:
IP: 192.168.100.157
MAC: 01-00-5e-7f-64-9d
NLB NIC is on the same VLAN & "Converged switch" in VMM as our mgmt/server traffic (That is to say it's on production VLAN, not on a separate vlan)
PROBLEM:
Can't ping 100.157. From VM guest itself, from host, or from Cisco 6509 switch.
Cisco show mac address lookup does not see that MAC anywhere
show ip igmp groups shows not igmp traffic at all. Clearing counters show sno multicast increment.
FURTHERMORE:
Host is setup thusly:
- Dell R810
- 8x1GbE Broadcom 5709c in a Server 2012 LACP/HASH team built via VMM powershell cmdlets
- On the physical switch side, those 8 nics are in a Cisco port-channel, trunked, all VLANs allowed
- Host has no "physical" nics per se, as in a 2008 R2 hyper-v host. Instead Host has these:
Set-VMNetworkAdapter -ManagementOS -Name "Live Migrate" -MinimumBandwidthWeight 35
Set-VMNetworkAdapter -ManagementOS -Name "MGMT" -MinimumBandwidthWeight 25
Set-VMNetworkAdapter -ManagementOS -Name "CSV" -MinimumBandwidthWeight 40
Set-VMNetworkAdapter -ManagementOS -Name "iSCSI #1" -MinimumBandwidthWeight 0
Set-VMNetworkAdapter -ManagementOS -Name "iSCSI #2" -MinimumBandwidthWeight 0
Set-VMNetworkAdapter -ManagementOS -Name "Aux" -MinimumBandwidthWeight 0
Get-VMSwitch outputs this on the converged v-switch:
ComputerName : My-host
Name : My awesome switch
Id : e2377ce3-12b4-4243-9f51-e14a21f91844
Notes :
SwitchType : External
AllowManagementOS : True
NetAdapterInterfaceDescription : Microsoft Network Adapter Multiplexor
Driver
AvailableVMQueues : 0
NumberVmqAllocated : 0
IovEnabled : False
IovVirtualFunctionCount : 0
IovVirtualFunctionsInUse : 0
IovQueuePairCount : 0
IovQueuePairsInUse : 0
AvailableIPSecSA : 0
NumberIPSecSAAllocated : 0
BandwidthPercentage : 0
BandwidthReservationMode : Weight
DefaultFlowMinimumBandwidthAbsolute : 0
DefaultFlowMinimumBandwidthWeight : 1
Extensions : {Microsoft NDIS Capture, Microsoft
Windows Filtering Platform, Microsoft
VMM DHCPv4 Server Switch Extension}
IovSupport : False
IovSupportReasons : {This network adapter does not support
SR-IOV.}
IsDeleted : False
Question:
Aside from a few of my favorite MS MVPs (shout out to
WorkingHardInIt for having this same question), I can't find much documentation on employing 2008 R2 NLB on guest VM within a fabric-oriented, VMM-built 2012 Hyper-Visor converged switch (no network virtualization...yet).
Yes I know all about VMM NLB but 1) I'm trying to wedge NLB in after building these VMs without a service template (NLB is the audible, essentially) and 2) MS NLB is configured in providers & I've created requisite VIP templates.
Even so, I ought to be able to create an NLB cluster without VMM's assistance in this scenario correct? Suboptimal, I know but possible, yes? Essentially I've put to synthetic NICs on each VM, set IPs manually, and assigned them to the same vlan. I can ping
each synthetic NIC, but not the cluster IP.
And yes: these particular vNICs have Mac Address Spoofing enabled.
Cisco:
I have a TAC case open with Cisco, but they can't quite figure it out either. IGMP Snooping enabled across the switch. And they insist that the old static arp entry to resolve this problem is no longer necessary, that Microsoft now complies with relevant
RFCs
Possible SOlution:
Only thing I can think of is flipping MulticastForwarding param below from disabled to enabled. Anybody ever tried it on a converged virtual switch on the Hyper visor? Is my virtual converged switch protecting
me from multicast igmp packets?
PS C:\utilities> Get-NetIPv4Protocol
DefaultHopLimit : 128
NeighborCacheLimit(Entries) : 1024
RouteCacheLimit(Entries) : 128
ReassemblyLimit(Bytes) : 1560173184
IcmpRedirects : Enabled
SourceRoutingBehavior : DontForward
DhcpMediaSense : Enabled
MediaSenseEventLog : Disabled
IGMPLevel : All
IGMPVersion : Version3
MulticastForwarding : Disabled
GroupForwardedFragments : Disabled
RandomizeIdentifiers : Enabled
AddressMaskReply : Disabled
Thanks for any thoughts.
RobertSorry for the poor follow-up Steven. We are using Server 2012 Hyper-V, not VMWare, on the hosts. You can close this but for the benefit of anyone who comes across it:
After working with Cisco, we decided not to implement multicast IGMP. Cisco says you still need to create a static ARP entry on the physical switch, though my cluster IP address & Microsoft NLB 2008 R2 were set up with igmp multicast, not multicast or
unicast. Here was his email:
Yes, we will need the static mapping for the NLB server in this case because the NLB mac address is multicast and the IP address is unicast. I was under the impression that even the server would be using IGMP but that’s not
the case. We won’t need to do the mapping for the nodes though if they use IGMP. To this end, following is the configuration that should make this work.rp 192.168.100.157
0100.5e7f.649d arpa
<u5:p></u5:p>
mac address-table static 0000.0000.649d vlan <> interface <> disable-snooping
ßThis is the switch interface where the NLB server is located<u5:p></u5:p>
interface vlan<>
<u5:p></u5:p>
ip pim sparse-dense-mode <- This is needed for the switch to elicit IGMP joins from the nodes<u5:p></u5:p>
end<u5:p></u5:p>
I don't think it got through to him that there was a virtual Layer 2/3 Hyper-V switch on top of 8 teamed GbE interfaces in LACP/hash. "Where the NLB server is located" = 1)a Cisco port-channel bound to one of six physical hosts; the NLB VM itself could be
on any of those port channels at any given time (We have a six node Hyper-V cluster).
Once I enabled pim I did see activity; but we killed this later as we realized we'd have to implement the same on 40+ managed routers globally
Anyway we further would have had to implement this across managed routers at 40 sites globally according to Cisco.
Robert -
Hyper-V Host displays error "Automatic Reapir Failed" and will not start after Power Outage
I have a single Dell PowerEdge configured as a Hyper-V host running Wink12 R2 Preview as a Native Boot VHD and hosting about a dozen VM's. We had a break in the power, just a lights on lights off, but it was enough to shut the machine down and trigger restart.
But the machine won't restart--instead it displays a message "Automatic repair Failed" and none of the Advanced repair options work to get the host back online.
Any assistance is greatly appreciated.
Michael J. MurphyHi MJMurphy,
Sorry for the delay , it seems likely a disk problem cause of unexpect interrupt .
Please boot the host from OS ISO and try to copy the VMs' VHD files and any important data to a safe place (such as external USB)via command prompt mode .
(perhaps you need to prepare raid driver if you have a raid controler )
Using diskpart --> list volume to check the status of disk partion or you can contact the hardware vendor to check it .
Hope this helps
Best Regards
Elton Ji
We
are trying to better understand customer views on social support experience, so your participation in this
interview project would be greatly appreciated if you have time.
Thanks for helping make community forums a great place. -
Can I have hyper V hosts connected to one domain then connect a VM to another domain.
I have three Quad port NICs- two will be connected to my DMZ domain and then to accommodate direct access I propose connecting one to my corporate LAN. I will then create the required vswitchs and place my direct access servers on DMZ switch and
the LAN switch for a 2 leg deployment?
Your thoughts please?Sure, no problem with that. Each OS whether physical host or guest VM can join any domain they can reach.
Mark B. Cooper, President and Founder of PKI Solutions Inc., former Microsoft Senior Engineer and subject matter expert for Microsoft Active Directory Certificate Services (ADCS). Known as “The PKI Guy” at Microsoft for 10 years. -
Hello
[Sorry for asking the same question in the wrong place/forum of "Hyper-V"]
I am using a CentOS 6.5 VM (Linux kernel 2.6.32-431) and (generation 2) CentOS 7 VM (Linux
kernel 3.10.0-123) with Hyper-V KVP daemon installed,
and I periodically query its IP (via using WMI or Powershell to query its KVP information) to manage it.
However, after querying its IP (KVP) for several times, its IP can’t
be queried or shown on Hyper-V Manager anymore (Windows VM is ok without this problem).
And here is the vmIntegrationService status of my CentOS 7 VM for the references.
PS C:\Users\Administrator> (Get-VM -name G2_CentOS7).vmIntegrationService
VMName Name Enabled PrimaryStatusDescription SecondaryStatusDescription
G2_CentOS7 Time Synchronization True OK
G2_CentOS7 Heartbeat True OK
G2_CentOS7 Key-Value Pair Exchange True OK The protocol version of the component installed ...
G2_CentOS7 Shutdown True OK
G2_CentOS7 VSS True No Contact
G2_CentOS7 Guest Service Interface False OK
I attached a simple KVP query Powershell script as
the follows, and this problem can be reproduced in couple minutes if you run two instances with this script at the same time.
$VMName = $args[0]
write-host "$VMName"
filter Import-CimXml
$CimXml = [Xml]$_
$CimObj = New-Object -TypeName System.Object
foreach ($CimProperty in $CimXml.SelectNodes("/INSTANCE/PROPERTY"))
if ($CimProperty.Name -eq "Name" -or $CimProperty.Name -eq "Data")
$CimObj | Add-Member -MemberType NoteProperty -Name $CimProperty.NAME -Value $CimProperty.VALUE
$CimObj
$CimObj = $null
for ($i=1 ; $i -le 10000 ; $i++) {
$a = Get-Date
write-host "$i - Time: " $a.ToLocalTime()
$vm = Get-WmiObject -Namespace root\virtualization\v2 -Query "Select * From Msvm_ComputerSystem Where ElementName='$VMName'"
$vm.ElementName
$vmkvp = Get-WmiObject -Namespace root\virtualization\v2 -Query "Associators of {$vm} Where AssocClass=Msvm_SystemDevice ResultClass=Msvm_KvpExchangeComponent"
$vmkvp.GuestIntrinsicExchangeItems | Import-CimXml
Actually, if your CentOS VM (has LIS) installed with KVP daemon running well,
my test script will show more than 4 keys (include NetworkAddressIPv4 or NetworkAddressIPv6 keys).
However, while the KVP daemon becomes problematic, it will only show a few keys
(ex. 4~6 keys) and at this moment, Hyper-V Manager also can't show IP address of it anymore and you may need to reboot the CentOS VM to recover it.
For example (KVP in 252 time is good, but KVP in 253 and 254 times become problematic)
252 - Time: 8/26/2014 7:19:42 PM
G2_CentOS7
localhost FullyQualifiedDomainName
3.1 IntegrationServicesVersion
10.1.145.190;192.168.122.1 NetworkAddressIPv4
fe80::215:5dff:fe91:b902 NetworkAddressIPv6
3.10.0-123.el7.x86_64 OSBuildNumber
0 OSDistributionData
0 OSDistributionName
199168 OSKernelVersion
7 OSMajorVersion
OSMinorVersion
CentOS Linux OSName
129 OSPlatformId
3.10.0 OSVersion
x86_64 ProcessorArchitecture
253 - Time: 8/26/2014 7:19:42 PM
G2_CentOS7
localhost FullyQualifiedDomainName
3.1 IntegrationServicesVersion
10.1.145.190;192.168.122.1 NetworkAddressIPv4
0 OSDistributionData
0 OSDistributionName
199168 OSKernelVersion
129 OSPlatformId
254 - Time: 8/26/2014 7:19:44 PM
G2_CentOS7
0 OSDistributionData
0 OSDistributionName
199168 OSKernelVersion
129 OSPlatformId
I
found the following patches and gave them a try, but the
problem still remains after applying these patches and the (generation 2) Ubuntu 14.04 with Linux kernel 3.13 also has this problem.
- Patch "Drivers:
hv: util: Fix a bug in the KVP code" has been added to the 3.14-stable tree
- Drivers: hv: vmbus: Fix a bug in the channel callback dispatch code
But (generation 1) Ubuntu 14.04 VM with Linux kernel 3.17 doesn't encounter this problem after querying its KVP/IP information on Hyper-V host several times.
Does anyone know what changes between Linux kernel 3.13 and 3.17 fix this issue?
Thanks,
PaulHi Dexuan and Mike,
Thanks for your help!
I did the patch with rpmbuild on CentOS 6.5 (with kernel 2.6.32-431).
However, the second patch cannot patch on CentOS6.5 since there is no target_cpu in such version.
So we are patching that with the following similar patch but this issue still occurs. (also replace ko in initrd)
https://lists.ubuntu.com/archives/kernel-team/2014-August/047725.html
Can you tell me if any patch I could miss or any suggestion?
Thanks for your time. -
Metric: SAP Host Agent Availability doubt
Hello folks!
I have a problem in technical monitoring from Solution Manager, and i would be happy, if someone could help me with this.
1 - SAP Solution Manager is trying to ping a IP that doesnt exists
2 - In the screenshot, the line is: "Ping to listener: xxx.xxx.xxx.xxx:port not successful".
Let's call the host of my managed object as "$HOST_NAME", and the IP as "$HOST_IP".
1 - The IP of $HOST_NAME is one, and the IP xxx.xxx.xxx.xxx that SolMan is trying to ping, is another, the ip doesnt even exists, because i can't reach him from anywhere. Is it not wrong? I think the "Listener" should be the $HOST_IP, not another...
2 - The correction for this problem, would be the adjustment of the "Listener" to the $HOST_IP? How could i do this?
Thanks alreadyHi,
Host agent runs on physical machine.
Can you goto agent admin and find host name and IP from there.
Divyanshu -
InterMedia Web Agent problems...
I have Oracle 8.1.5 and OAS 4.0.8 on an NT machine. I have installed interMedia Web Agent, having shut the OAS services down first, configured the application as described in the doc and restarted all services. I have one listener for OAS on port 90. When I try to access http://servername:90/intermedia/~admin I get a 'Page cannot be displayed' error. Why?
I have checked all the settings within OAS and they look fine. NT is running SP5. I can access http://servername:90/ fine.
Any help gratefully received!
LeeThe interMedia web agent is installed in the
OAS home and you can see the intermedia application and cartridge in the OAS admin page?
Have you checked the OAS wrb.log file? That is where errors should be displayed. If it is a web agent problem, check <windows_dir>wsclstch.err or log files in <OAS_HOME>/ord/web/logs.
What do they say about starting the agent?
Test the agent with http://host:port/intermedia/~test
Larry -
Poor MySQL Performance after updating Hyper-V Host to 2012R2
Hi
I recently updated one of our 2008R2 Hyper-V Hosts (Standalone) to 2012 R2. (64GB Memory, SSD Raid1 at LSI Controller + SAS RAID1 for Host OS)
Now one of the VMs (2008R2, Generation 1, fixed .vhdx on SSD RAID, 20GB fixed RAM, 4 Cores) performes really bad.
This VM is running a MySQL Database. The performance is VERY poor over a network as well as on the local system (Same task in our Application takes more than double the time as before).
The VM itself wasn't changed in any kind.
While file copy and overall performance of the VM seem to be OK, SQL queries are extremely slow.
CPU Usage of VM stays nearly idle, it only uses 2GB RAM, but the virtual Disk I/O only reaches about 8 KByte/s!!! (File Copy between 200 and 400 MB/s)
Checked same VM again on different Hardware -> same result...
Really have no ideas any more.Hi LiwoCN,
aha "my Application (Sage New Classic)".
Da kann ich bestimmt in deutsch helfen (otherwise I can translate it to english).
Sage New CL - das ist genau auch unser Ding. Hier kann ich (evtl.) ein paar Tipps geben:
Sofern es noch nicht getätigt ist, ist mit MySQL und Sage Classic Line und HYPER-V folgende Einstellung eklatant wichtig:
Sowohl den HYPER-V Host wie auch die Gast VM in den Energiesparoptionen auf "Höchstleistung" stellen. Sowie in den "Erweiterten Energieeinstellungen ändern" den Punkt "Prozessorenergieverwaltung" - Maximaler Leistungszustand des Prozessors" auf 100 % stellen.
Das ist zwar nicht im Sinne eines Energiesparens aber das treibt die Performanz der MySQL & New CL deutlich nach oben.
Fahren Sie dann in der New CL denn den DMOPERF Test. Optimal sind Werte unter 300. Alles drüber ist für den Nutzer quasi nicht akzeptabel (zu langsam). Wir erreichen mit HYPER-V recht optimale Werte um die 100 herum. Hinweis: Den DMOPERF Test am besten direkt
auf dem MySQL Server ausführen (also dort auch einen CL Clienten installieren). So grenzt man zumindest ein, ob es sich um ein VM Problem oder um ein LAN Problem handelt.
Des weiteren sollte man die VHD wo die MySQL DB drauf läuft als VHDX und als SCSI HDD einbinden. Sofern notwendig (abhängig von der Gast VM, wenn diese kleiner als Windows 2012 R2 ist), das Systemlaufwerk per IDE und das Datenlaufwerk (wo die MySQL drauf
läuft) als SCSI.
VMQ für die VM sollte eingeschaltet sein - aber Achtung nur richtige Server LAN Karten können dies! Alternativ kann man auch mit SR-IOV Treibern arbeiten. Aber wie gesagt, dass bringt nur Punkte, wenn die VM an sich ordentlich läuft (daher nicht primär wichtig).
Finetuning kann man dann noch in der sageclmy.ini (?:\Sage\Sage New Classic\MySQL 560\bin\) ausführen. Dies erfolgt aber auf eigenes Risiko, daher bitte vorsichtig mit den Parametern umgehen! Die Original sageclmy.ini zuvor bitte sichern!
Abhängig von dem VM Gast Werten (vor allem RAM), kann man die innodb Werte optimieren. Für eine VM mit 16 GB RAM haben wir diese Werte als optimal ermittelt.
innodb_data_file_path = ibdata1:10M:autoextend
innodb_buffer_pool_size = 10G
innodb_log_file_size = 2G
innodb_log_buffer_size = 8M
innodb_flush_log_at_trx_commit = 2
innodb_lock_wait_timeout = 50
innodb_log_files_in_group = 4
innodb_support_xa = 0
innodb_thread_concurrency = 0
myisam_max_sort_file_size = 4M
Obiges gibt vor allem dem Buffer Pool 10 GB RAM und den Log-Files 2 GB RAM. Die Sage Original Vorgaben sind deutlich geringer. Trotzdem immer bedenken, zu große Buffer machen sich bei vielen Schreibvorgängen auch negativ bemerkbar. Daher hier evtl. selbst
ein Optimum ermitteln.
Ich hoffe das hilft weiter...
Danke und liebe Grüße Oliver Richter -
Hi All,
I am facing a problem for SAP host agent. I have download the newest version of SAPHOSTAGENT. Decompression and run installation command.
<Path>saphostexec -install
Install service
Check Installation Path ...ok
Copy Executable ...
Service SAPHostExec is Stopped
Copying...ok
Generate Profile ...ok
Check Upgrade ...ok
Install windows service ...Service SAPHostExec registered!
ok
Check for saphostcontrol ...
Please enter password for user .\sapadm: ********
Register sapstartsrv: "C:\Program Files\SAP\hostctrl\exe\sapstartsrv.exe" -r -soap -p "C:\Program Files\SAP\hostctrl\exe\host_profile" -s SAP -n 99 -U .\sapadm -q -P ********
[ERROR] saphostcontrol registration failed wit exitcode -1
[ERROR] Failed
I have tried different version of SAPHOSTAGENT but still has the same error.
Please suggest.
RegardsHi,
Please follow SCN links SAP Host Agent - Frequently Asked Questions
Downloading and Installing Package SAPHOSTAGENT - SAP NetWeaver Landscape Virtualization Management, Standard - SAP Libr…
Regards,
Gaurav
Maybe you are looking for
-
Naming Services cannot work well!!!
Hi, I have configured the AM2005Q4 and Policy agent with apache, apache http.conf file is like ProxyRequests Off <Proxy *> Order deny,allow Allow from all </Proxy> ProxyPass /hzycportal http://exchange.hzliqun.com:8013/hzycportal ProxyPassReverse /hz
-
I LOVE my Verizon service.....
I am becoming depressed by the conflicts that I am seeing all over the VZN community. People are beginning to realize that the superior service coverage is coming at an excessively high cost (and not just in dollars). My company (26 years now) has ch
-
How do I get the brightness icon to go away on my MacBook?
How do I get the brightness icon to go away on my MacBook?
-
How to make fields Read only on all tabs in one go with the help of feeder classes in SAP EHS?
I have a requirement where once we press SAVE button i have to make all the fields read only in tabs highlighted below.Which is the best method to do it.? There are number of fields in all these tabs and we want to make them read only on particular A
-
when i start typing in address bare it start to search or predict the site or texts and i unchecked it from options but it still working i dont like it i want to cancel this option