Who Should Install This Hotfix?
This is a hotfix for customers running XenServer 6.0.2.
- Any issued driver disk and the Driver Development Kit (DDK) for XenServer 6.0.2 must be updated to be compatible with this hotfix. See CTX138920 – Driver Disks for XenServer 6.0.2 with Hotfix XS602E028 for a list of the affected driver disks that must be updated.
- After applying the hotfix to all hosts in a pool, customers should update the required driver disks before rebooting the XenServer hosts.
Issues Resolved In This Hotfix
This hotfix resolves the following issues:
- Attempts to generate crash dumps on Dell R720 servers can result in a XenServer host crash.
- The Control Domain (dom0) can experience CPU Soft Lockup issues, which may cause a XenServer host to freeze or reboot with the following message: BUG: soft lockup - CPU#0 stuck for 61s!.
- The kernel TCP stack can run the retransmission timer even though there are no packets to retransmit. When this happens, the warning message WARN_ON(tcp_write_queue_empty(sk)); will be displayed followed by a kernel crash.
- In some cases, GRO packets to Virtual Machines (VMs) may cause the XenServer host to crash.
- Some device drivers restrict their DMA capability to 32-bit resulting in the unnecessary use of the software bounce buffer (SWIOTLB). Exhaustion of the SWIOTLB can result in a host becoming unresponsive.
- In very rare cases, plugging a Virtual Block Device (VBD) into a VM can cause a race condition which leads to a host crash.
- In some cases, VMs appear to take a very long time (30 minutes or more) to shut down or reboot.
- Some utilities (for example xentop) display incorrect sector read/write count statistics.
- Timeouts can prevent the discovery of LUNs attached to Fibre Channel HBAs.
- Fibre Channel HBAs that take more than 10 seconds to establish a connection with the Control Domain (dom0), will not work as expected, as dom0 removes the Fibre Channel nodes after a 10-second timeout.
- New blkfront drivers which have enabled barrier support can cause blkfront to return errors to the guest kernel. This eventually causes the guest filesystems to become Read-only.
- Removed spurious error message in /var/log/messages: vhd-util: libvhd::vhd_validate_footer: invalid footer cookie.
- Some Linux VMs may erroneously send network packets greater than or equal to 64KiB causing network traffic to cease.
- Enabled all the ebtables modules in the dom0 kernel.
In addition, this hotfix also includes the following previously released hotfixes.
- CTX132823 – Hotfix XS602E001 - For XenServer 6.0.2
- CTX133166 – Hotfix XS602E003 - For XenServer 6.0.2
- CTX133812 – Hotfix XS602E005 - For XenServer 6.0.2
- CTX134479 – Hotfix XS602E007 - For XenServer 6.0.2
- CTX135303 – Hotfix XS602E011 - For XenServer 6.0.2
- CTX135623 – Hotfix XS602E013 - For XenServer 6.0.2
- CTX136479 – Hotfix XS602E021 - For XenServer 6.0.2
- CTX138631 – Hotfix XS602E027 - For XenServer 6.0.2
Installing the Hotfix
Customers should use either XenCenter or the XenServer Command Line Interface (CLI) to install this update. Once the update has installed, the server must be restarted for it to take effect. As with any software update, please back up your data before applying this hotfix. Citrix recommends updating all hosts within a pool sequentially. Upgrading of hosts should be scheduled to minimize the amount of time the pool runs in a "mixed state" where some hosts have been upgraded and some have not. Running a mixed pool of updated and non-updated hosts for general operation is not supported.
Note: The attachment to this article is a zip file. It contains both the hotfix update package, and the source code for any modified open source components. The source code is not necessary for hotfix installation: it is provided to fulfil licensing obligations.
Installing the update using XenCenter
- Download the update to a known location on a computer that has XenCenter installed.
- In XenCenter, on the Tools menu, select Install Update. This displays the Install Update wizard.
- Click Next to start the Wizard.
- Click Add to upload a new update.
- Browse to the location where you downloaded the hotfix, select it, and then click Open.
- From the list of updates select XS602E028.xsupdate and then click Next.
- Select the hosts you wish to apply the hotfix to, and then click Next.
- Follow the recommendations to resolve any upgrade prechecks and then click Next.
- Choose how to perform post-update tasks. In the Post update options section, select automatically or manually, and then click Install update.
- When the installation process is complete, click Finish to exit the wizard.
- Customers who select to manually perform the post-update tasks, must ensure to do so after installing the hotfix.
- Customers who select to automatically perform the post-update tasks should note that XenCenter reboots each host sequentially starting with the Pool Master, where possible VMs will be migrated to other running hosts to avoid VM downtime. When the Pool Master is being rebooted, XenCenter will be unable to monitor the pool.
Installing the update using the xe Command Line Interface
- Download the update file to a known location.
- Extract the xsupdate file from the zip.
- Upload the xsupdate file to the Pool Master by entering the following commands:
(Where hostname is the Pool Master's IP address or DNS name.)
xe patch-upload -sXenServer assigns the update file a UUID which this command prints. Note the UUID.
<hostname>-u root -pw
- Apply the hotfix to all hosts in the pool, specifying the UUID of the hotfix:
<hostname>-u root -pw
- Verify that the update was applied by using the patch-list command.
xe patch-list -sIf the update has been successful, the hosts field will contain the UUIDs of the hosts this patch was successfully applied to. This should be a complete list of all hosts in the pool.
<hostname>-u root -pw
- To verify in XenCenter that the update has been applied correctly, select the Pool, and then click the General tab. This displays the Pool properties. In the Updates section, ensure that the update is listed as Fully Applied.
- The hotfix is applied to all hosts in the pool, but it will not take effect until each host has been rebooted. For each host, migrate the VMs that you wish to keep running, and shutdown the remaining VMs before rebooting the host.
|Hotfix File md5sum||0ae16f16c30b5465a1b04499b80a7fc3|
|Hotfix Source Filename||XS602E028-src-pkgs.tar.bz2|
|Hotfix Source File md5sum||485fa19e14f89d9c1f7c81bb34993077|
|Hotfix Zip Filename||XS602E028.zip|
|Hotfix Zip File md5sum||e1663cc6c1d441937036b9ba4aab09e5|