nbdkit-scan-filter - scan disk prefetching data ahead of sequential reads
nbdkit --filter=scan PLUGIN [scan-ahead=false] [scan-clock=false]
[scan-forever=true] [scan-size=]NN nbdkit --filter=scan --filter=cache PLUGIN nbdkit --filter=scan --filter=cow PLUGIN cow-on-cache=truenbdkit-scan-filter is a filter that scans the disk prefetching data. It is sometimes useful if you expect that the client will read the disk sequentially.
The basic operation of the filter is that when a client connects, the filter will start issuing .cache (prefetch) requests to the plugin across the whole disk. Plugins which support this command will prefetch the data, making subsequent reads faster. For plugins which do not support this command, you can inject nbdkit-cache-filter(1) below (after) this filter, giving approximately the same effect. nbdkit-cow-filter(1) can be used instead of nbdkit-cache-filter, if you add the cow-on-cache=true option.
Various parameters can be used to tune scanning, although the defaults should be suitable in most cases.
A similar filter is nbdkit-readahead-filter(1).
In a number of significant cases this filter will do nothing. The filter will print a warning message if this happens.
For example nbdkit-curl-plugin(1) only supports serialize_requests, and so this filter cannot perform prefetches in parallel with the read requests.
The current filter only scans while there is at least one client connected.
The current filter only scans the default export and ignores all clients connecting to the non-default export name.
*We may be able to lift these restrictions in future.
.cache (prefetch)Very many plugins do not have the concept of prefetching and/or do not implement the .cache callback, and so there is no way for this filter to issue prefetches.
You can usually get around this by adding --filter=cache after this filter as explained above.
In particular if you use this filter together with nbdkit-cache-filter(1) or nbdkit-cow-filter(1), they will cache the whole content of the plugin into a temporary file. This may be many gigabytes of data, consuming all space in /var/tmp. Of course this is the whole point of using this filter, but you should be aware of it.
If using the cache filter, the total size of the cache can be limited (see "CACHE MAXIMUM SIZE" in nbdkit-cache-filter(1)).
By default the filter tries to stay ahead of incoming read requests. That is to say, it starts prefetching at the beginning of the disk and continues incrementally, but if the client issues a read beyond the current prefetch point then the filter skips forward and begins prefetching after the read.
However if you set this parameter to false, then this behaviour is disabled. The filter simply prefetches sequentially regardless of client requests.
By default, if all clients disconnect and then another client connects, prefetching resumes at the same place in the disk. (Like stopping and starting a clock.)
If you set this parameter to false, then the filter starts prefetching from the beginning of the disk again.
By default the filter scans over the disk once and then stops.
If you set this parameter to true, then after the disk has been prefetched completely, the filter goes back to the beginning and starts over, repeating this for as long as nbdkit is running and there are clients connected.
This parameter controls the prefetch block size. The default is 2M. This must be a power of 2 and most plugins will have their own limits on the amount of data they can prefetch in a single request.
The filter.
Use nbdkit --dump-config to find the location of $filterdir.
nbdkit-scan-filter first appeared in nbdkit 1.32.
nbdkit(1), nbdkit-cache-filter(1), nbdkit-cow-filter(1), nbdkit-file-plugin(1), nbdkit-readahead-filter(1), nbdkit-retry-filter(1), nbdkit-torrent-plugin(1), nbdkit-vddk-plugin(1), nbdkit-filter(3), qemu-img(1).
Richard W.M. Jones
Copyright Red Hat
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
Neither the name of Red Hat nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY RED HAT AND CONTRIBUTORS ''AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL RED HAT OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.