Display I/O statistics for the entire node
The io_s input request to the mmpmon command allows the system administrator to collect I/O statistics for the entire node.
The io_s (I/O statistics) request returns strings containing I/O statistics taken over all mounted file systems as seen by that node, and are presented as total values for the entire node. The values are cumulative since the file systems were mounted or since the last reset, whichever is most recent. When a file system is unmounted, its statistics are lost and its contribution to the total node statistics vanishes. Read and write statistics are recorded separately.
Table 1 describes the keywords for the io_s response, in the order that they appear in the output. These keywords are used only when the mmpmon command is invoked with the -p flag.
| Keyword | Description |
|---|---|
| _n_ | IP address of the node responding. This is the address by which GPFS knows the node. |
| _nn_ | The hostname that corresponds to the IP address (the _n_ value). |
| _rc_ | Indicates the status of the operation. |
| _t_ | Indicates the current time of day in seconds (absolute seconds since Epoch (1970)). |
| _tu_ | Microseconds part of the current time of day. |
| _br_ | Total number of bytes that are read from both disk and cache. |
| _bw_ | Total number of bytes that are written to both disk and cache. |
| _oc_ | Count of open() call requests that are serviced by GPFS. The open count also includes creat() call counts. |
| _cc_ | Number of close() call requests that are serviced by GPFS. |
| _rdc_ | Number of application read requests that are serviced by GPFS. |
| _wc_ | Number of application write requests that are serviced by GPFS. |
| _dir_ | Number of readdir() call requests that are serviced by GPFS. |
| _iu_ | Number of inode updates to disk, which includes inodes flushed to disk because of access time updates. |