Processing monitor data
The mon_procd daemon writes process data to the z/VM® monitor stream.
The data includes summary information and information of each process for up to 100 processes currently being managed by an instance of Linux® on z/VM to the z/VM monitor stream.
At the time of the sample interval, one sample monitor record is written for system summary data, then one sample monitor record is written for each process for up to 100 processes currently being managed by the Linux instance. If more than 100 processes exist in a Linux instance at a given time, processes are sorted by the sum of CPU and memory usage percentage values and only the top 100 processes' data is written to the z/VM monitor stream.
The monitor data in each record begins with a header (a time stamp, the length of the data, and the offset). The data after the header depends on the field "record number" of the 16-bit product ID and can be summary data or process data. See Reading the monitor data for details. The following is the format of system summary data passed to the z/VM monitor stream.
| Type | Name | Description |
|---|---|---|
| __u64 | time_stamp | Time at which the process data was sampled. |
| __u16 | data_len | Length of data following the header. |
| __u16 | data_offset | Offset from start of the header to the start of the process data. |
| __u64 | uptime | Uptime of the Linux instance. |
| __u32 | users | Number of users on the Linux instance. |
| char[6] | loadavg_1 | Load average over the last one minute. |
| char[6] | loadavg_5 | Load average over the last five minutes. |
| char[6] | loadavg_15 | Load average over the last 15 minutes. |
| __u32 | task_total | total number of tasks on the Linux instance. |
| __u32 | task_running | Number of running tasks. |
| __u32 | task_sleeping | Number of sleeping tasks. |
| __u32 | task_stopped | Number of stopped tasks. |
| __u32 | task_zombie | Number of zombie tasks. |
| __u32 | num_cpus | Number of CPUs. |
| __u16 | puser | A number representing (100 * percentage of total CPU time used for normal processes executing in user mode). |
| __u16 | pnice | A number representing (100 * percentage of total CPU time used for niced processes executing in user mode). |
| __u16 | psystem | A number representing (100 * percentage of total CPU time used for processes executing in kernel mode). |
| __u16 | pidle | A number representing (100 * percentage of total CPU idle time). |
| __u16 | piowait | A number representing (100 * percentage of total CPU time used for I/O wait). |
| __u16 | pirq | A number representing (100 * percentage of total CPU time used for interrupts). |
| __u16 | psoftirq | A number representing (100 * percentage of total CPU time used for softirqs). |
| __u16 | psteal | A number representing (100 * percentage of total CPU time spent in stealing). |
| __u64 | mem_total | Total memory in KB. |
| __u64 | mem_used | Used memory in KB. |
| __u64 | mem_free | Free memory in KB. |
| __u64 | mem_buffers | Memory in buffer cache in KB. |
| __u64 | mem_pgpgin | Data read from disk in KB. |
| __u64 | mem_pgpgout | Data written to disk in KB |
| __u64 | swap_total | Total swap memory in KB. |
| __u64 | swap_used | Used swap memory in KB. |
| __u64 | swap_free | Free swap memory in KB. |
| __u64 | swap_cached | Cached swap memory in KB. |
| __u64 | swap_pswpin | Pages swapped in. |
| __u64 | swap_pswpout | Pages swapped out. |
The following is the format of a process information data passed to the z/VM monitor stream.
| Type | Name | Description |
|---|---|---|
| __u64 | time_stamp | Time at which the process data was sampled. |
| __u16 | data_len | Length of data following the header. |
| __u16 | data_offset | Offset from start of the header to the start of the process data. |
| __u32 | pid | ID of the process. |
| __u32 | ppid | ID of the process parent. |
| __u32 | euid | Effective user ID of the process owner. |
| __u16 | tty | Device number of the controlling terminal or 0. |
| __s16 | priority | Priority of the process |
| __s16 | nice | Nice value of the process. |
| __u32 | processor | Last used processor. |
| __u16 | pcpu | A number representing (100 * percentage of the elapsed cpu time used by the process since last sampling). |
| __u16 | pmem | A number representing (100 * percentage of physical memory used by the process). |
| __u64 | total_time | Total cpu time the process has used. |
| __u64 | ctotal_time | Total cpu time the process and its dead children has used. |
| __u64 | size | Total virtual memory used by the task in KB. |
| __u64 | swap | Swapped out portion of the virtual memory in KB. |
| __u64 | resident | Non-swapped physical memory used by the task in KB. |
| __u64 | trs | Physical memory devoted to executable code in KB. |
| __u64 | drs | Physical memory devoted to other than executable code in KB. |
| __u64 | share | Shared memory used by the task in KB. |
| __u64 | dt | Dirty page count. |
| __u64 | maj_flt | Number of major page faults occurred for the process. |
| char | state | Status of the process. |
| __u32 | flags | The process current scheduling flags. |
| __u16 | ruser_len | Length of real user name of the process owner and should not be larger than 64. |
| char[ruser_len] | ruser | Real user name of the process owner. If the name is longer than 64, the name is truncated to the length 64. |
| __u16 | euser_len | Length of effective user name of the process owner and should not be larger than 64. |
| char[euser_len] | euser | Effective user name of the process owner. If the name is longer than 64, the name is truncated to the length 64. |
| __u16 | egroup_len | Length of effective group name of the process owner and should not be larger than 64. |
| char [egroup_len] | egroup | Effective group name of the process owner. If the name is longer than 64, the name is truncated to the length 64. |
| __u16 | wchan_len | Length of sleeping in function's name and should not be larger than 64. |
| char[wchan_len] | wchan_name | Name of sleeping in function or '-'. If the name is longer than 64, the name is truncated to the length 64. |
| __u16 | cmd_len | Length of command name or program name used to start the process and should not be larger than 64. |
| char[cmd_len] | cmd | Command or program name used to start the process. If the name is longer than 64, the name is truncated to the length 64. |
| __u16 | cmd_line_len | Length of command line used to start the process and should not be larger than 1024. |
| char [cmd_line_len] | cmd_line | Command line used to start the process. If the name is longer than 1024, the name is truncated to the length 1024. |
Use the time_stamp to correlate all process information that were sampled in a given interval.