postprocessing the solr documents are now not completely retrieved.
instead, only fiels, needed for the postprocessing are extracted. When
Solr document are written, this is done using partial updates.
This increases postprocessing speed by about 50% for embedded Solr
configurations. For external Solr configurations the enhancement should
be much higher because the postprocessing with remote Solr is very slow.
When doing partial updates to a remote Solr, this method should perform
much better than before, it is expected that this is even much higher
than the increase with local Solr.
(this eventually can benefit image search by using mime only)
reduce redundant field assignment for Solrdocuments created from URIMetadataNode (URIMetadataNode = SolrDocument with partially assigned fields)
it is now possible to get the results in two steps:
- first retrieve all IDs as given for a query
- then retieve each document individually
This was necessary for very large result sets where a query may run for
hours and is possibly terminated by a solr-internal timeout. This occurs
regulary during postprocessing and therefore this commit may fix
unwanted postprocessing terminations.
during document parsing; instead use the same references that would also
be written into the webgraph. That should cause that the webgraph and
the citation index express the exact same semantic.
see http://mantis.tokeek.de/view.php?id=437
test result (concurrency=7)
2000 docs = eom always
1000 docs = eom always
100 docs = eom never
chosen -> 200 docs (eom not encountered during test with 1GB mem setting)
formulated as edismax query but this was not set as query attribut. The
defType=edismax property needs a qf-field, so this was added as well. Do
not remove that field again! This fixes also a problem with title-unique
computation.
attribute in the <a> tag for each crawl. This introduces a lot of
changes because it extends the usage of the AnchorURL Object type which
now also has a different toString method that the underlying
DigestURL.toString. It is therefore not advised to use .toString at all
for urls, just just toNormalform(false) instead.
- Web servers may now deliver YaCy-specific http header field with a
title and keywords. The new http header fields are:
X-YaCy-Media-Title - to be used for media (image, audio, video) titles
X-YaCy-Media-Keywords - to be used for media (image, audio, video)
keywords
- both fields are written to document fields title and keywords and are
searched also during image search.
- to make the usage of arbitrary http header fields (including this new
fields) possible in the /api/push_p.json servlet, a new POST argument is
also introduced to push http header fields. The new POST attribute is
named "responseHeader-X" (where X is the counter). It is allowed to use
this attribute as multi-attribute several times, each can be filled with
a http header line.
- see /api/push_p.html for examples
- unique-postprocessing was destroying results from other
postprocessings; removed cross-updates as they had been not necessary
- unique-postprocessing did not restrict on same protocol
- inefficient concurrent update cache was redesigned completely
- increased limits for concurrent blocking queues to prevent early
time-out
This organizes all urls to be loaded in separate queues for each host.
Each host separates the crawl depth into it's own queue. The primary
rule for urls taken from any queue is, that the crawl depth is minimal.
This produces a crawl depth which is identical to the clickdepth.
Furthermorem the crawl is able to create a much better balancing over
all hosts which is fair to all hosts that are in the queue.
This process will create a very large number of files for wide crawls in
the QUEUES folder: for each host a directory, for each crawl depth a
file inside the directory. A crawl with maxdepth = 4 will be able to
create 10.000s of files. To be able to use that many file readers, it
was necessary to implement a new index data structure which opens the
file only if an access is wanted (OnDemandOpenFileIndex). The usage of
such on-demand file reader shall prevent that the number of file
pointers is over the system limit, which is usually about 10.000 open
files. Some parts of YaCy had to be adopted to handle the crawl depth
number correctly. The logging and the IndexCreateQueues servlet had to
be adopted to show the crawl queues differently, because the host name
is attached to the port on the host to differentiate between http,
https, and ftp services.
stack on html tag objects, not using a recursive parse-again method
which may cause bad performance and huge memory allocation. The new
method also produced better parsed image objects with exact anchor text
references.
- added order option to solr queries to be able to retrieve document
lists in specific order, here: link length
- added HyperlinkEdge class which manages the link structure
- integrated the HyperlinkEdge class into clickdepth computation
- extended the linkstructure.json servlet to show also the clickdepth
and other statistic information
a document. This is the upper limit for the clickdepth_i value which may
be shorter in case that the crawler did not take the shortest path to
the document.
instead of TreeMaps)
- enhanced memory footprint of database indexes (by introduction of
optimize calls)
- optimize calls shrink the amount of used memory for index sets if they
are not changed afterwards any more
different from normal requests. This happens if the remote solr is
actually a solrCloud; in such cases the luke request returns only the
result of the single solr peer, not the whole cloud.
also done: some refactoring.
a dublin core record inside of surrogate input files may now contain
tokens within the namespace 'md' (short for: metadata). The token names
must be valid withing the namespace of the solr field names. All
md-tokens inside of surrogate files then overwrite values within solr
documents before they are written to the solr index. This makes it
possible to assign collection names to each surrogate entry and also
ranking information can be added. Please see the example file.
work around the unfolding process in Solr's BinaryResponseWriter.
This was a huge performance bottleneck in the embedded solr connector
and the problem is actually on Solr side, but we have now a workaround.
- This made it possible to abstract a high-performance index access
method which is implemented as method getDocumentListByParams. That
method is also implemented in the SolrServerConnector and provides a
very efficient access to a solr index if the index is embedded.
- a popular use of the document list retrieval is a result count which
can now also make use of the new method, via getDocumentCountByParams.
- enhanced the Error cache which now does not store error documents
within the ram cache if the document is also written to solr. When
documents are retrieved from the cache, they are partly read from the
ram cache and if not existent there, from the Solr index.
Michael Peter Christen