Linux Process Scheduling Tool

Linux Process Scheduling Tool-schedutils implements parameter settings related to process scheduling through system calls, such as CPU affinity.
The schedutils software package includes taskset, irqset, lsrt, and CHRT. They can fully manage the process parameters.
Management.
1, taskset
Taskset [Options] [mask] [pid | Command [Arg]...]
Set/obtain the CPU binding of a specified process or run a new process based on the specified binding mask.
Mask: The binding relationship between a process and the CPU is specified by a 4-byte unsigned integer through a bit mask. Each binary represents
The lowest-end logic processor refers to the first logic processor. The highest-end logic processor refers to the last one. For example, when expressed in hexadecimal notation:
0x00000001 indicates the processor #0
0x00000003 indicates the processor #0 and #1
0 xffffffff indicates all processors (#0 to #31)
Option:
-P, -- PID operates on existing processes of the specified PID, rather than starting a new task.
-H, -- help displays help information.
-V, -- version displays version information.
Example:
Specify the binding mask to run the new command:
Taskset [mask] -- [command] [arguments]
Obtain the binding relationship of a specified process:
Taskset-P [pid]
Set CPU binding relationships for existing processes:
Taskset-P [mask] [pid]
2, irqset
Irqset [Options] [mask] [interrupt]
Chapter 2 Overview of advanced features
7
Set/obtain CPU binding for system interruption.
In SMP systems, system interruptions are randomly sent to any processor, and an interrupt can be bound by this tool.
Set to a group of processors. Only these processors can receive and handle the interrupt. Here, the mask is the same as the meaning in taskset. Parameter-free operation
The line irqset displays all interrupted binding relationships.
1.2 Functions
1.2.1 File System Support
Supports a variety of the latest log file systems, including XFS, JFS, reiserfs, and ext3.
For features and comparisons, see Chapter 2nd: Advanced file system guide.
1.2.2 iSCSI protocol supported
Provides further support for massive storage devices and related protocols, including iSCSI.
Iscsi allows the transmission of scsi I/O commands and data over a network using TCP/IP protocol, just like ing SCSI commands to optical fiber connections.
Parallel SCSI and SSA media. Install the iSCSI Device drive on the server side, receive the application I/O requests, and then
Use the iSCSI protocol to transmit them over the lan. The target storage device can be directly attached to the LAN, or you can use a router (protocol,
To connect to the LAN. Compared with other solutions, iSCSI has a small investment, outstanding Io performance, long transmission distance, and management and division.
Convenience.
Red Flag Advanced Server 4.0 supports iSCSI clients and does not support servers.
1.2.3 support for LVM and evms
Lv m (logical volume Management) greatly enhances the scalability and management efficiency of the disk subsystem.
Information, see Chapter 4th lv m User Manual.
Evms is an Enterprise Volume management system developed by IBM. It manages various volume management technologies, such as disk partitioning and Linux logical volume management.
(LV m), multi-disk (MD) management, os2 and Aix volume management and file system operations are unified in a single package.
For more information, see Chapter 5th: Use evms.
1.3 Reliability Improvement
The high-memory PTE patch is used to prevent the core from being down when the load is too high.
In the old kernel, Linux can only allocate the page table entry (page table entries, PTES) in the low-end memory.
1 GB limit. In applications such as Oracle 9ir2, a large amount of memory and processes are required, and the total space of Pte is large; connections at the same time
The number of users in the database is very large, and the core runtime may exceed the space of the PTE, even if the available idle memory and swap space system will
Chapter 2 Overview of advanced features
8
Suspended or down.
The high-memory PTE patch allows VMS to use the "high-end memory" shared pool for the allocation of Pte. When more and more users connect to the database
When an additional process is generated, the region for storing the PTE overflows to the high-end memory, which allows the system to support 3 more than the old kernel.
To 5 times the data.
Read Red Flag Advanced Server 4.0 release overview, which includes red