9

I am working on a powercli script to loop through each datastore in our environment and reclaim space on the thin provisioned LUNs. All LUNs in the environment are Dell Equal Logic.

vCenter version: 5.5.0_1750787

ESXi version: Dell 5.5-1331820(A01)

The problem I am running into is that exactly 30 minutes after executing the unmap command in the script I get the following error(s):

From PowerCLI:

Message:  An error occurred while communicating with the remote host:
InnerText:
At \\SOMEIP\Share\folder\vmware_unmap_datastore.ps1:96 char:9
+        $RETVAL = $ESXCLI.storage.vmfs.unmap(200, $ds.Name, $null)
+        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    + CatagoryInfo          : OperationsStopped: (:) [], HostCommunication
    + FullyQualifiedErrorId : VMware.VimAutomation.ViCore.Types.V1.ErrorHandling.HostCommunication

From: ESXi Host vpxa.log:

==> vpxa.log <==
2014-06-30T14:39:51.907Z [7265EB70 error 'vpxavpxaVmomi' opID=60e7b2a5-11]
[VpxaClientAdapter::InvokeCommon] Got exception while invoking executeSoap
on vmodl.reflect.ManagedMethodExecuter:ha-managed-method-executer:
'Operation timed out', backtrace:

--> backtrace[00] rip 13546113
Vmacore::System::Stacktrace::CaptureWork(unsigned int)

--> backtrace[01] rip 1339ea1b
Vmacore::System::SystemFactoryImpl::CreateQuickBacktrace(Vmacore::Ref<Vmacor
e::System::Backtrace>&)

--> backtrace[02] rip 1333fe15 Vmacore::Throwable::Throwable(std::string
const&)

--> backtrace[03] rip 133e7055
Vmacore::Http::UserAgentImpl::AsyncSendRequestHelper::WaitForResponse(long
long, Vmacore::Ref<Vmacore::Http::ClientResponse>&)

--> backtrace[04] rip 133e7596
Vmacore::Http::UserAgentImpl::TimedWaitSendRequest(Vmacore::Http::PendingReq
uestItem*, long long, Vmacore::Ref<Vmacore::Http::ClientResponse>&)

--> backtrace[05] rip 1397276c
Vmomi::SoapStubAdapterImpl::InvokeCommon(Vmomi::MoRef*,
Vmomi::ManagedMethod*, Vmacore::RefVector<Vmomi::Any> const&,
Vmomi::RequestContext*, bool, Vmacore::Ref<Vmomi::Any>&)

--> backtrace[06] rip 13972d78
Vmomi::SoapStubAdapterImpl::Invoke_Task(Vmomi::MoRef*,
Vmomi::ManagedMethod*, Vmacore::RefVector<Vmomi::Any>&,
Vmomi::RequestContext*, Vmacore::Ref<Vmomi::Any>&)

--> backtrace[07] rip 1285a104 /usr/lib/vmware/vpxa/bin/vpxa(+0x245104)
[0x1285a104]

--> backtrace[08] rip 1285a478 /usr/lib/vmware/vpxa/bin/vpxa(+0x245478)
[0x1285a478]

--> backtrace[09] rip 1399f764
Vmomi::StubImpl::_Invoke_Task(Vmomi::ManagedMethod*,
Vmacore::RefVector<Vmomi::Any>&, Vmacore::Ref<Vmomi::Any>&)

--> backtrace[10] rip 13a944e2
Vmomi::Reflect::ManagedMethodExecuterStub::ExecuteSoap(std::string const&,
std::string const&, std::string const&,
Vmomi::DataArray<Vmomi::Reflect::ManagedMethodExecuter::SoapArgument>*,
Vmacore::Ref<Vmomi::Reflect::ManagedMethodExecuter::SoapResult>&)

--> backtrace[11] rip 13a96a54 /lib/libvmomi.so(+0x2c2a54) [0x13a96a54]

--> backtrace[12] rip 1394d2c7
Vmomi::ManagedMethodImpl::Dispatch(Vmomi::ManagedObject*,
Vmacore::RefVector<Vmomi::Any>&, Vmacore::Ref<Vmomi::Any>&)

--> backtrace[13] rip 1394e312
Vmomi::ManagedMethodImpl::Invoke(Vmomi::ManagedObject*,
Vmacore::RefVector<Vmomi::Any>&, Vmacore::Ref<Vmomi::Any>&)

--> backtrace[14] rip 129f05ea /usr/lib/vmware/vpxa/bin/vpxa(+0x3db5ea)
[0x129f05ea]

--> backtrace[15] rip 12a95bab /usr/lib/vmware/vpxa/bin/vpxa(+0x480bab)
[0x12a95bab]

--> backtrace[16] rip 12858f22 /usr/lib/vmware/vpxa/bin/vpxa(+0x243f22)
[0x12858f22]

--> backtrace[17] rip 12a7458d VpxLRO::PreAndBodyStartWork(VpxLRO*)

--> backtrace[18] rip 12a67e41 /usr/lib/vmware/vpxa/bin/vpxa(+0x452e41)
[0x12a67e41]

--> backtrace[19] rip 12a715c5 /usr/lib/vmware/vpxa/bin/vpxa(+0x45c5c5)
[0x12a715c5]

--> backtrace[20] rip 12a76822 VpxLRO::LroMainEntry(VpxLRO*)

--> backtrace[21] rip 12a680a1 /usr/lib/vmware/vpxa/bin/vpxa(+0x4530a1)
[0x12a680a1]

--> backtrace[22] rip 12a714c8 /usr/lib/vmware/vpxa/bin/vpxa(+0x45c4c8)
[0x12a714c8]

--> backtrace[23] rip 12a87df2 VpxLroList::ThreadMainEntry()

--> backtrace[24] rip 1282691b
boost::detail::function::void_function_invoker0<void (*)(),
void>::invoke(boost::detail::function::function_buffer&)

--> backtrace[25] rip 134a76ca
Vmacore::System::ThreadPoolAsio::DoWork(Vmacore::Functor<void, Vmacore::Nil,
Vmacore::Nil, Vmacore::Nil, Vmacore::Nil, Vmacore::Nil, Vmacore::Nil,
Vmacore::Nil> const&, bool)

--> backtrace[26] rip 134a7f08
Vmacore::System::ThreadPoolAsio::DoOrEnqueueWork(Vmacore::System::ThreadPool
Asio::WorkTypeInfo&, Vmacore::Functor<void, Vmacore::Nil, Vmacore::Nil,
Vmacore::Nil, Vmacore::Nil, Vmacore::Nil, Vmacore::Nil, Vmacore::Nil>
const&, int)

--> backtrace[27] rip 134a8170
Vmacore::System::ThreadPoolAsio::HandleWork(Vmacore::System::ThreadPoolAsio:
:WorkType, Vmacore::Functor<void, Vmacore::Nil, Vmacore::Nil, Vmacore::Nil,
Vmacore::Nil, Vmacore::Nil, Vmacore::Nil, Vmacore::Nil> const&)

--> backtrace[28] rip 134b0421
Vmacore::System::ThreadPoolAsioWorkBinder::InvokeWork_0(Vmacore::System::Thr
eadPoolAsio::WorkType, boost::function<void ()> const&)

--> backtrace[29] rip 134abe7a
boost::detail::function::void_function_obj_invoker0<boost::_bi::bind_t<void,
void (*)(Vmacore::System::ThreadPoolAsio::WorkType, boost::function<void ()>
const&),
boost::_bi::list2<boost::_bi::value<Vmacore::System::ThreadPoolAsio::WorkTyp
e>, boost::_bi::value<boost::function<void ()> > > >,
void>::invoke(boost::detail::function::function_buffer&)

--> backtrace[30] rip 134b1efc
boost::asio::detail::completion_handler<boost::function<void ()>
> ::do_complete(boost::asio::detail::task_io_service*,
boost::asio::detail::task_io_service_operation*, boost::system::error_code,
unsigned int)

--> backtrace[31] rip 134af20e
boost::asio::detail::task_io_service::do_one(boost::asio::detail::scoped_loc
k<boost::asio::detail::posix_mutex>&,
boost::asio::detail::task_io_service::idle_thread_info*)

--> backtrace[32] rip 134a6e0c
Vmacore::System::ThreadPoolAsio::RunWorkerThread()

--> backtrace[33] rip 134abed3
boost::detail::function::void_function_obj_invoker0<boost::_bi::bind_t<void,
boost::_mfi::mf0<void, Vmacore::System::ThreadPoolAsio>,
boost::_bi::list1<boost::_bi::value<Vmacore::Ref<Vmacore::System::ThreadPool
Asio> > > >, void>::invoke(boost::detail::function::function_buffer&)

--> backtrace[34] rip 1354fc94
Vmacore::System::ThreadPosixPrivate::ThreadBegin(void*)

--> backtrace[35] rip 12fd4d6a /lib/libpthread.so.0(+0x6d6a) [0x12fd4d6a]

--> backtrace[36] rip 17bcde3e /lib/libc.so.6(clone+0x5e) [0x17bcde3e]

-->

2014-06-30T14:39:51.914Z [7265EB70 info 'commonvpxLro' opID=60e7b2a5-11]
[VpxLRO] -- FINISH task-internal-241280 --  --
vmodl.reflect.ManagedMethodExecuter.executeSoap --

2014-06-30T14:39:51.914Z [7265EB70 info 'Default' opID=60e7b2a5-11] [VpxLRO]
-- ERROR task-internal-241280 --  --
vmodl.reflect.ManagedMethodExecuter.executeSoap:
vmodl.fault.HostCommunication:

--> Result:

--> (vmodl.fault.HostCommunication) {

-->    dynamicType = <unset>,

-->    faultCause = (vmodl.MethodFault) null,

-->    msg = "",

--> }

--> Args:

-->

2014-06-30T14:39:51.914Z [72367B70 error
'SoapAdapter.HTTPService.HttpConnection'] Failed to read header on stream
<io_obj p:0x720a3e30, h:18, <TCP '0.0.0.0:0'>, <TCP '0.0.0.0:0'>>:
N7Vmacore15SystemExceptionE(Connection reset by peer)

I have been unable to find a 30 minute session timeout which I can tweak in the script. I attempted to remove or increase the Web Session Timeout in PowerCLI without success:

Set-PowerCLIConfiguration -WebOperationTimeoutSeconds -1 -Scope Session -Confirm:$false
Set-PowerCLIConfiguration -WebOperationTimeoutSeconds 3600 -Scope Session -Confirm:$false

Any direction or advice on improving the following script would be very helpful:

Clear-Host
$ErrorActionPreference = "Inquire"
$DebugPreference = "Continue"
$VerbosePreference = "Continue"

@"
## vmware_unmap_datastore.ps1 #################################################
Usage:        powershell -ExecutionPolicy Bypass -File ./vmware_unmap_datastore.ps1

Purpose:      Dumps Datastore (in GB): Capacity, Free, and Uncommitted space to
              to CSV and runs ESXCli command 'unmap' to retrieve unused space
              on Thin Provisioned LUNs.

Requirements: Windows Powershell and VI Toolkit

Assumptions:  All ESXi hosts have access to all datastores

TO DO:        Import Dell Equal Logic Module, get Used space before/after unmap

History:      06/20/2014  -  Created
###############################################################################
"@

## Prompt Administrator for vCenter Server ####################################
###############################################################################
[System.Reflection.Assembly]::LoadWithPartialName('Microsoft.VisualBasic') | Out-Null
$VCServer = [Microsoft.VisualBasic.Interaction]::InputBox(
                "vCenter Server FQDN or IP",
                "PowerCLI Prompt: vCenter Server Query",
                "VCENTERHOST.SUB2.SUM1.DOMAIN.COM")


## Filename and path to save the CSV ##########################################
###############################################################################
$timestamp = $(((get-date).ToUniversalTime()).ToString("yyyyMMdd"))
$output_path = [Environment]::GetFolderPath("mydocuments")
$output_file = $output_path + "\datastore_info-" + $timestamp + ".csv"

## Ensure VMware Automation Core Snap In is loaded ############################
###############################################################################
if ((Get-PSSnapin -Name VMware.VimAutomation.Core -ErrorAction SilentlyContinue) -eq $null) {
     Add-PSSnapin VMware.VimAutomation.Core      }

## Unmap can take hour+ per data store on first run, remove timeout ###########
###############################################################################
Set-PowerCLIConfiguration -WebOperationTimeoutSeconds -1 -Scope Session -Confirm:$false

## Ignore Certificates Warning ################################################
###############################################################################
Set-PowerCLIConfiguration -InvalidCertificateAction Ignore -Scope Session -Confirm:$false

## Connect to vCenter Server ##################################################
# Prompt user for vCenter creds every time unless creds are stored using:
# New-VICredentialStoreItem -Host $VIServer -User "AD\user" -Password 'pass'
###############################################################################
$VC = Connect-VIServer $VCServer
Write-verbose "Connected to '$($VC.Name):$($VC.port)' as '$($VC.User)'"

## Connect to first ESXi host in list to run unmap ESXCLI #####################
###############################################################################
$ESXiHost = Get-VMHost |Select-Object -first 1
$ESXCLI = Get-EsxCli -VMHost $ESXiHost
Write-Verbose "Using ESXi host '($ESXiHost)' for CLI"


## Establish structure to store CSV data ######################################
# Try to open a CSV file, if it doesn't exist a new one will be created.
###############################################################################
try {
    $report = @( Import-Csv $output_file )
    }
catch {
    $report = @()
    }

## CSV Collect Data ###########################################################
# Function to collect datastore usage information to be stored in CSV
###############################################################################
function get_datastore_usage {
    Write-Verbose "[ $($dsv.Name) ] - Gathering statistics..."
    $row = "" |select TIMESTAMP, DATASTORE, CAPACITY_GB, FREE_GB, UNCOMMITED_GB
    $row.TIMESTAMP = $(((get-date).ToUniversalTime()).ToString("yyyyMMddThhmmssZ"))
    $row.DATASTORE = $ds.Name
    $row.CAPACITY_GB = [int]($ds.CapacityGB)
    $row.FREE_GB = [int]($ds.FreeSpaceGB)
    $row.UNCOMMITED_GB = [int]($dsv.Summary.Uncommitted / (1024 * 1024 * 1024))
    return $row
    }

## Unmap ######################################################################
# unmap creates a maximum of 200 (changable) 1MB files at a time to 100%.
###############################################################################
function reclaim_datastore_used_space {
    Write-Verbose "[ $($dsv.Name) ] - Running unmap, can take 30+ minutes"
    try {
        $RETVAL = $ESXCLI.storage.vmfs.unmap(200, $ds.Name, $null)
        }
    catch [VMware.VimAutomation.Sdk.Types.V1.ErrorHandling.VimException.ViError]{
        Write-Verbose $_.Exception.Message -foregroundcolor magenta
        }
    }

## Loop through datastores ####################################################
# Loops through all datastores seen by vCenter.  If the datastore is accessible
# and capable of thinprovisioning: Gathers datastore usage data, runs unmap
###############################################################################
foreach ($ds in Get-Datastore) {
    $dsv = $ds |Get-View
    if ($dsv.Summary.accessible -and $dsv.Capability.PerFileThinProvisioningSupported) {
        Write-Verbose "[ $($dsv.Name) ] - Refreshing Datastore Data..."
        $dsv.RefreshDatastore()
        $dsv.RefreshDatastoreStorageInfo()
        $report += get_datastore_usage
        reclaim_datastore_used_space
        }
    }


## Write CSV data to file #####################################################
###############################################################################
$report |Export-Csv $output_file -NoTypeInformation

## Open CSV file using Notepad ################################################
###############################################################################
Start-Process notepad -ArgumentList $output_file

## Properly disconnect from vCenter Server ####################################
###############################################################################
Disconnect-VIServer $VC -Confirm:$false

Note: The unmap command continues to run in the background, the blocks can be seen being unmapped in the ESXi Host hostd.log:

Unmap:  Async Unmapped 200 blocks from volume ...
Lars
  • 198
  • 1
  • 10
  • Not a solution, but a band-aid... If I bump the `unmap` blocks from the default 200 to 800, my test LUN (4.5 TB) finishes in under 30 minutes. Dell/Equal-logic recommends using 200. – Lars Jun 30 '14 at 16:03
  • The '800' only worked on LUNs where unmap was run recently to reclaim space. 30 minute timeout was still hit on LUNs where space had not be reclaimed in months. – Lars Jul 02 '14 at 13:21
  • I have had no luck on resolving the 30 minute SOAP timeout. Anybody have any ideas of where else to look? thanks, – Lars Jul 09 '14 at 17:44

3 Answers3

1

You mentioned:

I have been unable to find a 30 minute session timeout which I can tweak in the script. I attempted to remove or increase the Web Session Timeout in PowerCLI without success:>

Please try this, as it looks like the timeout from the vcenter host itself.

vpxd.cfg file

To configure the timeout and maximum number of sessions:

Open the vpxd.cfg file using a text editor.

The vpxd.cfg file is located at: Windows XP: C:\Documents and Settings\All Users\Application Data\VMware\VMware VirtualCenter\vpxd.cfg Windows Server 2003: C:\Documents and Settings\All Users\Application Data\VMware\VMware VirtualCenter\vpxd.cfg Windows Server 2008: C:\ProgramData\VMware\VMware VirtualCenter\vpxd.cfg

Locate the tag within the file. Set the maxSessionCount and sessionTimeout parameters.

For example:

... 500 30

Note: The session timeout value is specified in minutes. In vCenter Server 2.5.x, the default values are 30 minutes and 100 sessions. You also may need to restart the VMware VirtualCenter Server service after making this change.

Jon Conley
  • 328
  • 1
  • 7
  • 1
    thank you for your response. The page you linked has a warning saying to not apply this fix to vSphere 5+. I am on 5.5. I checked the vpxd file anyway and there is not a subset under . `Warning: Do not apply this article if you are using vSphere 5.0 or later. The default session limit in vSphere 5.x is 2000. Implementing the change below in vSphere 5.x reduces the default SOAP connections allowed by vCenter Server by 1500.`. I will attempt to add the section using the vSphere 5 connection default, and the higher timeout and post my findings. r/ – Lars Jul 18 '14 at 19:54
  • I'm looking in our environment, too. I see the setting for 30 minutes in web.xml under C:\Program Files\VMware\Infrastructure\tomcat\webapps\vws\WEB-INF\web.xml Looks like you might have to bounce that app after making that change. 30 – Jon Conley Jul 18 '14 at 20:10
  • I made the change listed in your 2004663 link, and ran the script. Strangely the operation timed out at 25 minutes instead of 30, with a timeout of 120 minutes set in /var/lib/vmware/vsphere-client/webclient.properties. I will try again Monday in case it was an oddity. I will also try your latest suggestion to web.xml. Thanks, – Lars Jul 18 '14 at 20:37
  • Hopefully you got the auto award of the bounty? The other answer I wasn't able to down-vote (he suggests something ALREADY in the script that doesn't work) due to not having 125 points. the 25 minute timeout yesterday was due to someone restarting vCenter under me. Re-running the last attempt shortly. – Lars Jul 21 '14 at 12:49
  • I changed the web.xml session timeout on the vcenter server from 30 to 120, restarted the server. The ESXCli timeout still occurs after 30 minutes. – Lars Jul 21 '14 at 14:42
0

Try this

Set-PowerCLIConfiguration -Scope Session -WebOperationTimeoutSeconds -1

Credit

MichelZ
  • 11,008
  • 4
  • 30
  • 58
  • The Web Operation Timeout you suggested is already in the script. It changed it from the default '300' seconds (5 minutes) to supposedly removing the timeout altogether. I also tried it with '3600' seconds (1 hour), it failed at 30 minutes into the ESXcli operation. Monday I will dig into the ESXi host's logs and hopefully dig out something useful. – Lars Jun 28 '14 at 14:37
0

I had the same problem. Set the timeout first as mentioned before in this thread.

Set-PowerCLIConfiguration -WebOperationTimeoutSeconds -1

Then connect to the host directly if possible.

$h = Connect-VIServer "hostname"

$ESXCLI = Get-EsxCli -VMHost "hostname" -server $h

That did it for me. In my case it looked like it was the vCenter that did the timeout, not my powercli.

Cory Knutson
  • 1,866
  • 12
  • 20