NAME

memx − memory exerciser

SYNTAX

/usr/field/memx [ −h ] [ −s ] [ −ofile ] [ −ti ] [ −mj ] [ −pk ]

DESCRIPTION

The memx memory exerciser spawns processes to exercise memory by writing and reading three patterns: 1’s and 0’s, 0’s and 1’s, and a random pattern.

You specify the number of processes to spawn and the size of memory to be tested by each process. The first process is a shared memory exerciser, the remaining are standard memory exercisers. The memx exerciser will run until the process receives a <CTRLC> or a kill -15 pid. 

A logfile is made in /usr/field for you to examine and then remove. If there are errors in the logfile, check the /usr/adm/syserr/syserr.<hostname> file, where the driver and kernel error messages are saved.

OPTIONS

The memx options are:

−hPrint the help message for the memx command.

−sDisable shared memory testing. 

−ofileSave diagnostic output in file. 

−tiRun time in minutes (i).  The default is to run until the process receives a <CTRLC> or a kill -15 pid. 

−mjThe memory size in bytes (j) to be tested by each spawned process.  Must be greater than 4095.  The default is (total-memory)/20. 

−pkThe number of processes to spawn (k).  The default is 20.  The maximum is also 20. 

RESTRICTIONS

The memx exerciser is restricted by the size of swap space available.  The size of the swap space and the size of internal memory available will determine how many processes can run on the system.  For example, If there were 16Mbytes of swap space and 16Mbytes of memory, all of the swap space would be used if all 20 spawned memory exercisers were running.  In that event, no new processes would be able to run. On systems with large amounts of memory and small swap space restricting the number of memory exercisers and/or size of memory being test must be done.

If there is a need to run a system exerciser over an NFS link or on a diskless system there are some restrictions.  For exercisers that need to write into a file system, such as fsx(,), the target file system must be writable by root.  Also the directory, in which any of the exercisers are executed, must be writable by root because temporary files are written into the current directory.  These latter restrictions are sometimes difficult to overcome because often NFS file systems are mounted in a way that prevents root from writing into them.  Some of the restrictions may be overcome by copying the exerciser to another directory and then executing it.

EXAMPLES

The following example tests all of memory by running 20 spawned processes until a <CTRLC> or kill -15 pid is received. 

% /usr/field/memx

The following example runs 10 spawned processes, memory size 500,000 bytes, for 180 minutes in the background.

% /usr/field/memx -t180 -m500000 -p10 &

SEE ALSO

Guide to System Exercisers

Maintenance