zCX Constraint Relief
IBM® z/OS® Container Extensions (IBM zCX) makes it possible to integrate Linux on Z software applications as Docker Containers within z/OS without requiring a separate Linux server. Application developers can develop and data centers can operate popular open source packages, Linux applications, IBM software, and third-party software together with z/OS applications and data. The ability to colocate the Linux and z/OS parts of a workload improves operational control, leverages z/OS Qualities of Service, and typically improves performance. The overall improvement is possible because data can be more easily moved between the zCX address space hosting Linux and the address spaces hosting the z/OS parts.
IBM has made multiple enhancements to zCX so it can host larger workloads. The enhancements improve the scalability of the zCX address space and also grow the resources that it can access.
The maximum size of a disk that zCX can use has been increased to 1 TB. The previous limit was 46 GB, so this is a significant enhancement. You can create a larger data disk using the provision workflow, or you can add one to an existing instance by using the add data disks workflow. The larger disk size gives you the flexibility to have a single large disk or a pool of multiple smaller disks.
This support is included in APAR OA60920.
In a zCX instance, user data disks are used to store data such as Docker containers and the contents of the home directory. The number of user data disks that you can allocate and associate with a zCX instance has been increased to 245. With the increase, you can leverage both the large disk size capability and the increased number of disks to run very large workloads in zCX. The data disks that zCX provides to the Linux guest are combined into a single logical volume.
The increased number of disks can be utilized in several ways. When the system programmer creates an instance, the zOSMF provision workflow will allow you to define up to 10 data disks. If the instance requires more space later, the add data disks workflow can be run to add up to 10 more data disks. This process can be repeated as necessary until the instance has at most 245 data disks.
This support is included in APAR OA60452.
More Memory for More Containers
The maximum amount of memory that can be allocated to a zCX instance has been increased to 1 TB (1024 GB). This is a 10x increase over the previous limit of 100 GB. The memory limit for an instance is defined during provisioning and can be modified using the reconfigure zOSMF workflow. In order to exploit this memory, the maximum number of simultaneously running Docker containers has been raised to 1000.
Using a high memory limit enables you to run a larger number of containers or run containers that have very large memory requirements. If you assign a large amount of memory to a zCX instance, you need to ensure that the z/OS system has enough memory for normal operation. All memory assigned to zCX is page-fixed, meaning that it is unavailable for any other purpose.
If you will be using large amounts of memory in a zCX instance, you should consider configuring the instance to use 1 M (large) or 2 G (huge) pages (APAR OA59573). The larger page sizes reduce the page table overhead and can significantly improve startup and processing times.
This support is included in APARs OA59864 and OA59943.
Consolidated Version Message
A new message has been added to zCX that displays the versions of each of its constituent components. These components are:
The new consolidated version message is displayed during startup of a zCX instance as message
This message shows the version information for each of the above components. It also displays the version
number of the available appliance, which helps identify whether an appliance is downlevel. You can display
this message from a running zCX instance with the following console command, where < instname > is the name of the zCX instance:
MODIFY < instname >,DISPLAY,VERSION
An example message is as follows:
GLZB022I zCX instance ZCXANGIO version informationBootloader: HZDC7C0 oa599343.12.0 1.2.1Current Appliance: HZDC7C0 oa609233.17.0 1.16.020210217T195101ZAvailable Appliance: HZDC7C0 oa60923
The Bootloader line refers to the Linux bootloader version. When zCX service is applied, this may be updated. A simple restart of the zCX guest will pick up the new version, and it will be reflected here.
The Current Appliance line refers to the version of the Linux guest. This value remains constant until the appliance is upgraded to a more recent version. To upgrade an appliance version, ensure that zCX service is applied, and then run the zOSMF upgrade workflow. The workflow will walk you through the steps required to upgrade the appliance. When the upgrade is complete, a restart of the zCX instance will pick up the new version level. Note that the string ending in “Z” is an ISO 8601 timestamp and can be used to determine the build date of the Linux guest.
If you have a number of zCX instances running in z/OS, you can check the version numbers of each instance to determine whether they would benefit from an upgrade.
The Available Appliance line is similar to the previous line, but instead refers to the version of the appliance available in z/OS. When zCX service is applied to the system, this value will be updated. If this value is different than the Current Appliance value, it indicates that this zCX instance would benefit from an upgrade.
The Virtualization Layer line refers to the version level of the zCX hypervisor address space. This code is updated when zCX service is applied and is shared among all the running zCX instances on the system. To pick up the latest level of this code, you can restart any running zCX instances.
The message also indicates when the zCX instance was started, which can help to identify long-running instances for service purposes.
The Workflows Performed messages indicate what workflow operations (if any) were performed on this specific instance. For example, the above message shows that the provision workflow was run on March 26, 2021, by workflow version 1.0.36. No other workflows have yet been performed, as indicated by the text “N/A”.
This support is included in APARs OA58996 and OA59835.
zCX can now monitor the resource usage of the Linux guest within an instance. This monitoring covers the following areas:
If the utilization of one of those areas crosses a monitoring threshold, a message is printed to the job JESMSGLG. It identifies the affected resource and the level of urgency. A message is only issued when a threshold is crossed. If a resource continually remains at a given level, only a single message is printed.
These messages are intended to alert the z/OS system programmer that a particular zCX instance is suffering from resource starvation. If any of the resources are exhausted, the instance could unexpectedly terminate.
A set of resource alert messages is as follows:
GLZM011I zCX instance ZCXANGI1 alert: memory warning.Utilization at 80%.GLZM011I zCX instance ZCXANGI1 alert: memory critical.Utilization at 99%.GLZM011I zCX instance ZCXANGI1 alert: memory normal.
In addition, a message is printed to the job SYSPRINT that shows the utilization levels for each resource:
Mar 29 15:24:01 azd_resmon: crit MEM: 99 SWAP: 61 ROOT: 7 DATA: 65
To resolve a constrained resource situation, first identify the resource and then consult the following table:
This support is included in APAR OA60303.
zCX can now use an IPv6 address in addition to an IPv4 address for network communications. This capability enables the Linux guest to communicate over an IPv6 network as needed. An IPv4 address is still required.
To utilize this feature, set the IPv6 address in the zOSMF provision or reconfigure workflows. Make sure to modify the appropriate DVIPA settings in the z/OS TCPIP profile.
This support is included in APAR OA59508.
Anthony Giorgio is an Advisory Software Engineer with 20 years of mainframe experience. He currently works on the z/OS Container Extensions (zCX) team in Poughkeepsie, NY. Anthony holds a BS and MS in Computer Science from the NYU Tandon School of Engineering.
Debbie Noll contributed to the technical and editorial review of this article.