Content-type: text/html
Man page of presto
presto
Section: Maintenance Commands (8)
Index
Return to Main Contents
NAME
presto - Controls and monitors the Prestoserve file system accelerator
SYNOPSIS
/usr/sbin/presto
[flags]
DESCRIPTION
The presto command allows you to accelerate file systems,
obtain Prestoserve status, and administer Prestoserve.
If invoked with no flags, presto displays
the Prestoserve state (either UP, DOWN, or ERROR),
the number of bytes of nonvolatile memory the Prestoserve cache is using,
how long the cache has been enabled, the write cache efficiency, and
the state of the backup battery or batteries.
When the Prestoserve state is UP, Prestoserve
improves I/O performance to accelerated file systems
by caching synchronous disk write operations to nonvolatile memory.
When the Prestoserve state is DOWN,
all I/O requests are passed to the appropriate disks.
If it detects a disk error during a write back, Prestoserve
enters the ERROR state and disables itself. However, Prestoserve
continues to maintain the integrity of cached data. Some possible disk
error conditions are: the disk drive is write protected or off line,
a cable problem exists, or a bad disk block exists.
Also, if there is insufficient backup battery power,
Prestoserve will enter the ERROR state.
FLAGS
Disables Prestoserve and writes the Prestoserve cache data to the intended
disks. Only those file systems specified by
filesystem are disabled. You specify filesystem as a
directory mount point (for example, /usr).
-
Do not specify a block
device because some functional subsystems, such as the
Advanced File System (AdvFS), can map more than one block device to
a mount point.
-
If no file systems
are specified, all accelerated file systems are disabled, and the
Prestoserve state is set to DOWN.
-
This flag does not reset Prestoserve statistics.
-
The -d flag takes effect before the -u or -R
flags.
Similar to the -d flag, but sets the Prestoserve state to DOWN only if
the specified directory is the root of a mounted file system.
Otherwise, the following message is displayed:
-
presto: directory is not a file system root
Writes the Prestoserve cache data to the intended disks,
but leaves the cache data intact.
-
If the flag is used and the Prestoserve state is UP, then the cache
data is written to the intended disks, and the state remains UP.
If the Prestoserve state is
DOWN, then there is no data to write to the disks, and the state remains
DOWN. If the state is ERROR, then as much of the cache data as possible is
written to the intended disks.
-
Note that unlike the -R
flag, the data in the Prestoserve cache remains after it is written to
the intended disks.
Invokes the specified operations on the remote machine hostname by
using an RPC protocol. The remote machine must be running the
prestoctl_svc daemon to allow the use of the -h flag.
The prestoctl_svc -n flag is required to allow the use of
-h with the -s, -d, and -u administrative flags.
See prestoctl_svc(8) for details.
To avoid possible security and performance
problems, Digital strongly recommends against
using the -h option.
See the rsh(1) reference page for alternative methods
to use for remote operations.
Lists the accelerated file systems and their mount points in a format
similar to the mount command. This
flag can be used with either NFS client or server machines.
Lists all local and remote mounted file systems and their mount points
that have been accelerated. Any unusual
Prestoserve state for a file system is displayed after the mount point.
The unusual states include:
Instead of directly accessing the nonvolatile memory, the
file system's device receives the Prestoserve data
only after the data is first copied to main memory.
Prestoserve acceleration is not enabled on the file system.
An error occurred using the file system,
and the Prestoserve cache data has still not been written successfully
to the intended disks.
Displays Prestoserve information. The information includes
the current Prestoserve state; the statistics for write,
read, and total operations; and battery status. For example:
-
# /usr/sbin/presto -p
dirty = 52, clean = 7, inval = 0, active = 2
count hit rate clean hits dirty hits allocations passes
write: 1516 65% 0 989 511 15
read: 8 0% 0 0 0 8
total: 1524 65% 0 989 511 23
state = UP, size = 0x7e000 bytes
statistics interval: 00:00:13 (13 seconds)
write cache efficiency: 66%
All batteries are ok
-
The current Prestoserve statistics account for all Prestoserve buffers.
A dirty buffer contains a disk block image that has not been written
to disk. A clean buffer contains a valid disk block image that has been
written to disk. An inval buffer does not presently contain a disk block
image. An active buffer is currently in transition to
disk, meaning that a write operation has started but has not completed on
that buffer.
-
For each Prestoserve cache read or write operation, Prestoserve increments
a counter, as follows:
The clean hits counter shows the number of hits (block matches) on the clean
buffers.
The dirty hits counter shows the number of hits on the dirty buffers.
Each dirty hit on a write represents a physical
disk write that was avoided entirely, while a dirty
hit on a read represents the same condition for
a disk read.
The allocations counter shows the number of new buffers that had to be
allocated for disk block images.
The passes counter shows the number of I/O operations that Prestoserve
passed directly to the real device driver.
-
In addition, for each Prestoserve cache read or write operation, the
presto -p command displays the count,
which is the sum of the four counters
explained previously; the hit
rate percentage, which is the ratio of clean hits
and dirty hits to the total count and which indicates the
effectiveness of the Prestoserve cache; and the write cache efficiency
percentage, which is computed from the ratio of
write dirty hits to the number of writes copied into the Prestoserve cache.
-
The presto -p command also displays information about the Prestoserve
battery state. The command displays the battery state as ok,
low, or disabled.
-
Some processors support chargeable
batteries and use self tests to determine if a battery needs charging.
If you use the presto -p command on a machine that
supports chargeable batteries and is running Prestoserve locally, the
battery state can also be in self test or is charging.
Note that if you use the -p flag with the -h flag (or
if you use the dxpresto command,
batteries that are being self-tested or charged will
be displayed as disabled.
Writes as much of the Prestoserve cache data as possible to disk,
discards the data it could not write, purges all the Prestoserve
buffers, and sets the Prestoserve state to DOWN.
-
Unlike the -d flag, the -R
flag discards the Prestoserve cache data that could not be written to
disk and resets the statistics information.
The flag is useful when Prestoserve cache data
is not needed or if you cannot get Prestoserve out of the ERROR state.
Take care when using the -R flag, because it destroys
Prestoserve cache data.
-
The -R flag takes effect before the -u flag.
Sets the size of the Prestoserve cache to size
bytes. The size can be specified using the
decimal or hexadecimal conventions. For example, both 262144 and 0x40000
represent 256 Kbytes.
-
If the -s
flag is used and the current Prestoserve state is UP, the state
is set to DOWN, the Prestoserve cache is resized, and the state is set
to UP.
-
You may want to use the -s
flag to determine how Prestoserve performs with a reduced amount of
nonvolatile memory. Note that the size of the Prestoserve cache
cannot be larger than the default maximum size. If you specify
a size that is larger than the default maximum size, the
default maximum size is used.
Sets Prestoserve state to UP, and enables acceleration.
-
Only those file systems specified by filesystem are
enabled. You specify filesystem as a
directory mount point (for example, /usr).
-
Do not specify a block
device because some functional subsystems, such as the
Advanced File System (AdvFS), can map more than one block device to
a mount point.
-
If no file systems are specified, all local
writable file systems that are mounted
will have Prestoserve enabled. File systems that are presently
accelerated will remain accelerated.
-
If Prestoserve state was ERROR, Prestoserve attempts to write any blocks
that are in the cache to disk to ensure that
the previous error condition has been corrected.
Similar to the -u flag, but sets the Prestoserve state to UP only if
the specified directory is the root of a mounted file system.
Otherwise, the following message is displayed:
-
presto: directory is not a file system root
Specifies verbose mode. This flag prints extra information to standard
output. The information can be used for debugging purposes.
FILES
RELATED INFORMATION
prestoctl_svc(8), dxpresto(8X), presto(7),
prestotab(4), prestosetup(8)
Guide to Prestoserve
delim off
Index
- NAME
-
- SYNOPSIS
-
- DESCRIPTION
-
- FLAGS
-
- FILES
-
- RELATED INFORMATION
-
This document was created by
man2html,
using the manual pages.
Time: 02:40:31 GMT, October 02, 2010