Replacements for the vmtune and schedtune commands
The vmtune and schedtune commands were replaced by the vmo, ioo, and schedo commands. Both the vmo and ioo commands together replace vmtune, while the schedo command replaces schedtune. All existing parameters are used by the new commands.
The ioo command manages all the I/O-related tuning parameters, while the vmo command manages all the other Virtual Memory Manager, or VMM, parameters previously managed by the vmtune command. All three commands are part of the bos.perf.tune fileset, which also contains the tunsave, tunrestore, tuncheck, and tundefault commands. The bos.adt.samples fileset still includes the vmtune and schedtune commands, which are compatibility shell scripts calling the vmo, ioo, and schedo commands as appropriate. These compatibility scripts only support changes to parameters which can be changed interactively. Parameters that need bosboot and then require a reboot of the machine to be effective are no longer supported by the vmtune script. To change those parameters, users must now use the vmo -r command. The vmtune command options and parameters in question are as follows:
| The previous vmtune option | Usage | New command |
|---|---|---|
| -C 0|1 | page coloring | vmo -r -o pagecoloring=0|1 |
| -g n1 -L n2 | large page size number of large pages to reserve | vmo -r -o lgpg_size=n1 -o lgpg_regions=n2 |
| -v n | number of frames per memory pool | vmo -r -o framesets=n |
| -i n | interval for special data segment identifiers | vmo -r -o spec_dataseg_int=n |
| -V n | number of special data segment identifiers to reserve | vmo -r -o num_spec_dataseg=n |
| -y 0|1 | p690 memory affinity | vmo -r -o memory_affinity=0|1 |
The vmtune and schedtune compatibility
scripts do not ship with AIX®.
You can refer to the following tables to migrate your settings to
the new commands:
| The schedtune option | The schedo equivalent | Function |
|---|---|---|
| -a number | -o affinity_lim=number | Sets the number of context switches after which the SCHED_FIF02 policy no longer favors a thread. |
| -b number | -o idle_migration_barrier=number | Sets the idle migration barrier. |
| -c number | -o %usDelta=number | Controls the adjustment of the clock drift. |
| -d number | -o sched_D=number | Sets the factor used to decay CPU usage. |
| -e number | -o v_exempt_seconds=number | Sets the time before a recently suspended and resumed process is eligible for resuspension. |
| -f number | -o pacefork=number | Sets the number of clock ticks to delay before retrying a failed fork call. |
| -F number | -o fixed_pri_global=number | Keeps fixed priority threads in the global run queue. |
| -h number | -o v_repage_hi=number | Changes the system-wide criterion used to determine when process suspension begins and ends. |
| -m number | -o v_min_process=number | Sets the minimum multiprogramming level. |
| -p number | -o v_repage_proc=number | Changes the per process criterion used to determine which processes to suspend. |
| -r number | -o sched_R=number | Sets the rate at which to accumulate CPU usage. |
| -s number | -o maxspin=number | Sets the number of times to spin on a lock before sleeping. |
| -t number | -o timeslice=number | Sets the number of 10 ms time slices. |
| -w number | -o v_sec_wait=number | Sets the number of seconds to wait after thrashing ends before adding processes back into the mix. |
| The vmtune option | The vmo equivalent | The ioo equivalent | Function |
|---|---|---|---|
| -b number | -o numfsbuf=number | Sets the number of file system bufstructs. | |
| -B number | -o hd_pbuf_cnt=number | This parameter has been replaced with the pv_min_pbuf parameter. | |
| -c number | -o numclust=number | Sets the number of 16 KB clusters processed by write behind. | |
| -C 0|1 | -r -o pagecoloring=0|1 | Disables or enables page coloring for specific hardware platforms. | |
| -d 0|1 | -o deffps=0|1 | Turns deferred paging space allocation on and off. | |
| -e 0|1 | -o jfs_clread_enabled=0|1 | Controls whether JFS uses clustered reads on all files. | |
| -E 0|1 | -o jfs_use_read_lock=0|1 | Controls whether JFS uses a shared lock when reading from a file. | |
| -f number | -o minfree=number | Sets the number of frames on the free list. | |
| -F number | -o maxfree=number | Sets the number of frames on the free list at which to stop frame stealing. | |
| -g number | -o lgpg_size number | Sets the size, in bytes, of the hardware-supported large pages | |
| -H number | -o pgahd_scale_thresh=number | Sets the number of free pages in a mempool under which the system scales back read-ahead. | |
| -i number | -r -o spec_dataseg_int=number | Sets the interval to use when reserving the special data segment identifiers. | |
| -j number | -o j2_nPagesPerWriteBehindCluster= number | Sets the number of pages per write-behind cluster. | |
| -J number | -o j2_maxRandomWrite=number | Sets the random-write threshold count. | |
| -k number | -o npskill=number | Sets the number of paging space pages at which to begin killing processes. | |
| -l number | -o lrubucket=number | Sets the size of the least recently used page replacement bucket size. | |
| -L number | -o lgpg_regions=number | Sets the number of large pages to be reserved. | |
| -M number | -o maxpin=number | Sets the maximum percentage of real memory that can be pinned. | |
| -n number | -o nokilluid=number | Specifies the uid range of processes that should not be killed when paging space is low. | |
| -N number | -o pd_npages=number | Sets the number of pages that should be deleted in one chunk from RAM when a file is deleted. | |
| -p number | -o minperm%=number | Sets the point below which file pages are protected from the repage algorithm. | |
| -P number | -o maxperm%=number | Sets the point above which the page stealing algorithm steals only file pages. | |
| -q number | -o j2_minPageReadAhead=number | Sets the minimum number of pages to read ahead. | |
| -Q number | -o j2_maxPageReadAhead=number | Sets the maximum number of pages to read ahead. | |
| -r number | -o minpgahead=number | Sets the number of pages with which sequential read-ahead starts. | |
| -R number | -o maxpgahead=number | Sets the maximum number of pages to be read-ahead. | |
| -s 0|1 | -o sync_release_ilock=0|1 | Enables or disables the code that minimizes the time spent holding the inode lock during sync. | |
| -S 0|1 | -o v_pinshm=0|1 | Enables or disables the SHM_PIN flag on the shmget system call. | |
| -t number | -o maxclient%=number | Sets the point above which the page stealing algorithm steals only client file pages. | |
| -T number | -o pta_balance_threshold= number | Sets the point at which a new PTA segment is allocated. | |
| -u number | -o lvm_bufcnt=number | Sets the number of LVM buffers for raw physical I/Os. | |
| -v number | -r -o framesets=number | Sets the number of framesets per memory pool. | |
| -V number | -r -o num_spec_dataseg= number | Sets the number of special data segment identifiers to reserve | |
| -w number | -o npswarn=number | Sets the number of free paging space pages at which the SIGDANGER signal is sent to processes. | |
| -W number | -o maxrandwrt=number | Sets a threshold for random writes to accumulate in RAM before pages are synchronized to disk using a write-behind algorithm. | |
| -y 0|1 | -r -o memory_affinity=0|1 | This parameter does not exist. Memory affinity is always on if the hardware supports it. | |
| -z number | -o j2_nRandomCluster=number | Sets random write threshold distance. | |
| -Z number | -o j2_nBufferPerPagerDevice= number | Sets the number of buffers per pager device. |