VMware ESX 3.5 Update 5, Patch ESX350-200911201-UG: Updates VMkernel, Service Console, hostd
Article ID: 329262
Updated On:
Products
VMware vSphere ESXi
Issue/Introduction
Release Date: December 3, 2009
| Download Size: 306 MB Download Filename: ESX350-200911201-UG.zip md5sum: 1ca07f9947ba57e7d9a37610094f0bbc |
|
Resolution
Summaries and Symptoms
This patch fixes the following issues or adds the following enhancements for ESX 3.5 Update 5:
- Adds new options for the esxcfg-vswitch command. These options are -X and -x , and let you to set and get the value for maxActive uplinks respectively.
- In the case of excessive serial logging in large PCPU systems, ESX Server hosts fail with messages on a purple screen. This release of ESX Server modifies the behavior of serial logging of ESX Server hosts. By default serial logging is enabled. In VirtualCenter Server, the logSynchronous option is removed. This option is accessible from VirtualCenter > Configuration > Advanced Settings > VMkernel -> Boot -> logSynchronous.
You can enable or disable serial logging as follows:
• From VirtualCenter Server - Set the option Misc.LogToSerial to an appropriate value. This option can be accessed from VirtualCenter > Configuration > Advanced Settings > Misc.
• From ESX Server - Run the esxcfg-advcfg –s <value> /Misc/LogToSerial command from the service console.
- Resolves an issue of poor performance of the virtual machines for a short duration immediately after vMotion of virtual machines between ESX Server hosts. The following scenarios might lead to the occurrence of this issue:
- A guest operating system is using large pages frequently.
- Virtual machines are already swapping before the start of vMotion.
- Fixes an issue where the vcbSnapAll command might fail with an error when the backup_destination is set as scp .
- The upgrade procedure fails with the More than one vmkernel found error when upgrading from ESX 3.5 to ESX 3.5 Update 2. This issue might occur when upgrading from ESX 3.5 GA to any ESX 3.5 update releases prior to ESX 3.5 Update 5.
- ESX machines hosting passive MSCS nodes report reservation conflicts during storage operations. For more information refer KB 1009287.
- ESX Server host might stop responding during shutdown, in a specific use case that arises, only if virtual machines on the host are configured for autostart or autostop. In this release, the script that handles autostart and autostop of virtual machines is fixed.
- Fixes an issue where virtual machines might power off during a snapshot operation when path failover occurs on the storage where virtual machines reside.
- When you repair or modify VMware Tools on a virtual machine from Control Panel > Add or Remove Program, the installer fails and rolls back with the Fatal Error during Installation error. In this release, the repair option from Control Panel > Add or Remove Program is removed. You must instead run setup.exe directly from the original media that was used for a fresh installation.
- The memory scrubber service in Solaris pops memory balloon by accessing the memory pages reserved by the balloon driver. In this release, the memory scrubber service is disabled when the balloon driver is loaded, thereby reducing the memory balloon pop rate.
- If VMware Tools is upgraded on a Linux guest operating system, the guest operating system loses network connectivity. After the upgrade, the guest operating system stops the network service and does not restart the service automatically. This issue exists even for an automatic upgrade of VMware Tools on Linux guest operating systems.
- Fixes an issue where the Virtual Machines tab in VirtualCenter displays Windows 2008 Server virtual machines as Windows Vista virtual machines.
- Starting with ESX 3.5 Update 5 release, VMware Tools installs a udev rule file on Linux operating systems with kernel version greater than or equal to 2.6.13. This rule file changes the default timeout value of VMware virtual disks to 180 seconds. This helps the guest operating system to survive better during a failover scenario.
- In ESX 3.5 Update 5, VMware Tools is updated with fixes and the version is incremented to 7.4.8. The fixes are as follows:
- SVGA driver fix for Windows guest operating systems.
- The auto-upgrade process completes without any manual intervention.
- The memory scrubber service is disabled when the balloon driver is loaded, thereby reducing the memory balloon pop rate.
- The Virtual Machines tab in VirtualCenter displays Windows 2008 Server virtual machines as Windows Vista virtual machines.
- If you uninstall VMware Tools or the VSS components of VMware Tools, applications that require the msvcp71.dll file fail to open when the virtual machine is rebooted.
- VMware Tools upgrade overwrites the SVGA driver without checking the version of the existing SVGA driver.
- Adds support for the esxcfg-nic -a command for BCM5715S NIC in ESX Server 3.5.
- If the / partition of ESX Server service console is full, any command that modifies esx.conf results in the loss of all configuration data. For example, when the file system is full, attempts to modify vswitch configuration result in loss of all configuration data. As a result, networking fails after reboot. This release of ESX Server fixes this issue preventing data loss under such a condition. Also, an error message mentioning that the filesystem is full is displayed.
- In earlier releases of ESX Server, the updated RDM size is displayed on the virtual machines only if the ESX Server is restarted. Starting with this release, the updated RDM size is displayed in the virtual machines by running the LUN rescan. You need not restart the ESX Server.
- In a CIB (Cluster-in-a-box) configuration, a failover might occur due to reservation conflicts arising from VMFS operations performed by another host sharing the same volume. When the conflict arises, SCSI errors are logged in the Windows event log.
- Fixes an issue where ESX Server fails with Exception 14 in world vsi_traverse when running the vm-support command.
- Fixes an issue where running multiple rescan of VMFS volumes on ESX Server machines that have IBM Director agents installed might result in an error on a purple screen.
- Guest operating systems can access SCSI CD-ROM. For more information, refer KB 1008673.
- As per ACPI specifications, data formats returned by the _PRS and _CRS methods must match for any interrupt link devices implementing them. If the BIOS incorrectly implements different formats, the ESX Server might fail to boot and display a purple screen. The fix detects inconsistency in data formats and lets the system boot ignoring the faulty information.
- ESX host fails if its available TCP/IP sockets are exhausted and an NFS Client has a directory mounted. This issue is fixed in a preceding patch release.
- In prior releases, when VMFS heartbeat corruption is encountered in the VMFS volumes, virtual machines might fail to power on. However, the expected messages indicating heartbeat corruption for the VMFS volume are not logged in the VMkernel logs. In ESX 3.5 Update 5, log messages similar to the following are logged in the VMkernel logs:
Mar 24 17:06:25 blr-2nd-1-dhcp274 vmkernel: 0:01:34:08.655 cpu4:1039)WARNING: FS3: ReadHostPulse:1017: FS 49c2390a-1976a3cb-1449-001aa0ae5b61 may be damaged.
Mar 24 17:06:25 blr-2nd-1-dhcp274 vmkernel: 0:01:34:08.666 cpu4:1039)WARNING: FS3: ReadHostPulse:1028: Corrupt heartbeat detected at offset 0x3b9e00:
[state 0 offset 0 gen 0 owner 00000000-00000000-0000-000000000000 journalAddr 0To resolve this issue, customers are recommended to contact VMware support.
- Includes a fix that provides a mechanism to improve the performance of ESX Server hosts which might be impacted due to IRQ sharing.
- In a configuration that contains multiple virtual machines with large memory sizes, memory reclamation is triggered on idle virtual machines even though there is free memory available.
- After the user unmaps a few LUNs from EMC Invista or EMC DGC storage arrays and rescans the LUNs on ESX hosts, the warning message: D:0x0/H:0x0 0x0 0x0 0x0 appears.
- When vMotion fails because of high memory-intensive load of virtual machines or any similar issue, the virtual machine might get powered off. This issue is resolved in this release, allowing virtual machines to continue functioning on the source if vMotion fails.
- Fixes an issue where a race condition in the DentryCache initialization causes the system to fail with an error on a purple screen.
- A guest operating system's memory usage might be overestimated on Intel systems that support EPT technology or on AMD systems that support RVI technology. This might cause the memory alarms in the VirtualCenter Server to go off even if the guest operating system is not actively accessing a lot of memory.
- Fixes file permission issues when reading swap on NFS.
- In earlier releases of ESX Server, you can mark the active storage paths OFF anytime by running the esxcfg-mpath command from the service console of the ESX Server host or through the VI Client or VirtualCenter Server. Starting with ESX 3. 5 Update 5, the active path cannot be marked OFF if there is active I/O operation on the target storage array or when there is active reservation on the storage path. However, the non-active paths may be marked OFF any time.
- Some Solaris virtual machines frequently reset the balloon driver. In this release, two parameters are introduced that lets you disable ballooning. These parameters are:
- sched.mem.balloon.resetlimit : Number of balloon resets after which the ballooning function is disabled. The default value is 12.
- sched.mem.balloon.decayperiod : Number of seconds after which the value of the reset counter is halved. The default value is 43200 seconds (12 hours).
These parameters can be configured in the .vmx file. When the sched.mem.balloon.resetlimit parameter becomes zero, ballooning is disabled. After ballooning is disabled on a virtual machine, you can enable it only by restarting the virtual machine.
- During an auto-upgrade of VMware Tools on a Windows virtual machine through the VI client, the installer might display a dialog box requesting user input. To finish the upgrade process, you must log in to the virtual machine and respond to the dialog box. In this release, the issue is fixed and the auto-upgrade process completes without any manual intervention.
- Fixes an issue where virtual machines using the Virtual Machine Interface (VMI) might stop responding.
- Fixes an issue where Java applications running in SMP (Symmetric Multi-Processing) virtual machines with VMI enabled might fail.
- Fixes an issue where Symmetric Multi-Processing (SMP) virtual machines utilizing AMD's Rapid Virtualization Indexing (RVI) feature might display a black screen up to 30 seconds after powering on.
- Fixes an issue where the virtual machines utilizing the Virtual Machine Interface (VMI) might boot very slowly.
- Fixes an issue where users can set NTP properties in UTC time zone by using VI Client after installing ESX Server.
- Fixes a buffer corruption issue that occurs in hostd when multiple clients simultaneously establish a connection with an ESX Server host.
- In this release, the mptscsi_2xx driver is updated to version 2.06.74 to provide a better support for shared SAS environments. The mptscsi_2xx driver supports products based on the HP Modular Smart Arrays (MSA) and many other serial attached SCSI (SAS) arrays.
- Fixes an issue where systems with Intel igb cards run in dynamic conservative mode by default and have a lower rate of interrupts per second. Also, the TCP throughput performance is optimum.
- When several multiple-NPIV configured virtual machines are started simultaneously, the virtual ports are not created for all virtual machines. This fixes resolves the inconsistency in the number of virtual ports created on the proc node.
- In ESX Server 3.5 Update 5, NetQueue is disabled by default in the NetXen driver. Currently, VMware does not support NetQueue on the NetXen driver.
- Corrects the Driver version numbers to be compliant with the standard RPM versioning scheme.
- If the bnx2 NIC is reset by NetWatchDog, bnx2 initialization fails. This release of ESX Server introduces a fix that stops resetting the NIC if it fails initialization five times by NetWatchDog and displays a message similar to the following:
Jul 15 10:17:09 prme-stg270 vmkernel: 0:00:12:51.090 cpu12:1127)<5>bnx2: giving up resetting vmnic1 initialization failed 4 times
- Starting with this release, ESX Server offers support for the following guest operating systems:
- Windows 7 Enterprise (32-bit and 64-bit)
- Windows 7 Ultimate (32-bit and 64-bit)
- Windows 7 Professional (32-bit and 64-bit)
- Windows 7 Home Premium (32-bit and 64-bit)
- Windows 2008 R2 Standard Edition (64-bit)
- Windows 2008 R2 Enterprise Edition (64-bit)
- Windows 2008 R2 Datacenter Edition (64-bit)
- Windows 2008 R2 Web Server (64-bit)
- Ubuntu Desktop 9.04 (32-bit and 64-bit)
- Ubuntu Server 9.04 (32-bit and 64-bit)
Feedback
Yes
No
Powered by
