jdk-based logger tend to block
at java.util.logging.Logger.log(Logger.java:476) in concurrent
environments. This makes logging a main performance issue. To overcome
this problem, this is a add-on to jdk logging to put log entries on a
concurrent message queue and log the messages one by one using a
separate process.
- FTPClient uses the concurrent logging instead of the log4j logger
references_internal_url_sxt because they had been shown to be
superfluous. The citation of referrer in the host browser is possible
without them. Therefore now the host browser does not only show
internal, but also external referrer to each link.
While the values for the reference evaluation are computed, also a
backlink-structure can be discovered and written to the index as well.
The host browser has been extended to show such backlinks to each
presented links. The host browser therefore can now show an information
where an document is linked. The new citation reference is computed as
likelyhood for a random click path with recursive usage of previously
computed likelyhood. This process is repeated until the likelyhood
converges to a specific number. This number is then normalized to a
ranking value CRn, 0<=CRn<=1. The value CRn can therefore be used to
rank popularity within intra-domain link structures.
- move setting of system property solr.directoryFactory=solr.MMapDirectoryFactory to solrcore.properties
- add check of os.arch for 64bit system, if it fails use default/solrcore.x86.properties (if exists) as solrcore.properties
reason: on 32bit MMapDirectoryFactory may fail with.....
Caused by: java.io.IOException: Map failed
at sun.nio.ch.FileChannelImpl.map(FileChannelImpl.java:849)
at org.apache.lucene.store.MMapDirectory.map(MMapDirectory.java:283)
soft commits, reduced caching size of search events, ensured that solr
results are processed before connection is closed to keep that stuff not
too long in RAM
API recording for this method so it can be repeated automatically. The
index dump generation is now also available for API recording. Added
some synchronization in backend which was necessary for this.
id to be tested, but with a collection of ids. This will cause only a
single call to solr instead of many. The result is a much better
performace when testing the existence of many urls. The effect should
cause very much less IO during index transmission, both on sender and
receiver side.
- removed httpclient 3.1 which has been used by solrj < 4.x.x and is now
not used any more
- fixed some parts in YaCy which used methods from httpclient 3.1
Because the index size is now provided by solr, and the only way to do
that is a match for [* TO *], a size computation is quite complex and
time-consuming. Therefore this patch prevents that the method is called
at all and if necessary puts a DOS-preventing barrier in front of it.
- check geolocation coordinates and accept only those, which are
well-formed
- the solr push process does not stop crawling any more if after 20
requests to Solr Solr does not accept the record. Instead, a severe log
entry asks the user to create a bug request
- intruduced raw-queries for the re-introduced byId-Queries (they are
hopefully faster than full edismax queries)
- removed the cached solr connector (testing this) to rely only on the
solr built-in search caches. That should save some RAM (also). We will
see if this is usable.
adjusted to smaller and 1-core devices.
- the workflow processor now starts no process at all. these are started
as soon as parser/condenser/indexing queues are filled.
- better abstraction
holds the number of documents for the host where the document is hosted.
This is necessary for ranking and the norming of references per local
host in the ranking computation.
references_external_i and references_exthosts_i. These can be used to
count and evaluate the number of external links to every web page. An
experimental ranking function can be i.e.:
div(add(references_internal_i,product(references_external_i,references_exthosts_i)),add(clickdepth_i,1))
- removed unused solr access classes
- made snippet generation for documents aus YaCy RWI/DHT concurrent (as
it was before the search process removation)
- reduced the number of remote results in settings file because the
processing of such mass documents add is too CPU-intensive (in Solr)
- removed 'worker' processes
- no internal time-out behaviour: methods either are successful or
return null
- waiting is only done on top-level
- removed snippet-production; this is replaced by solr snippets
- removed statistics based on solr size queries (they had been VERY
long); the statistics (like suggestions or tag cloud) are now again
based on the old but very fast RWI index. In portal or intranet mode the
RWI index is usually switched off; if you like to have statistics again
then you must switch on the rwis again in this mode.
- fixed many bugs regarding correct page counter
index (new solr core webgraph) .. this is now off by default
- completely redesigned this servlet
- added description how to attach a remote solr
- adjusted naming of servlet and menues
- moved 'lazy initialization' attribut from IndexSchema to
IndexFederated (this is a general option) back again.
The default schema uses only some of them and the resting search index
has now the following properties:
- webgraph size will have about 40 times as much entries as default
index
- the complete index size will increase and may be about the double size
of current amount
As testing showed, not much indexing performance is lost. The default
index will be smaller (moved fields out of it); thus searching
can be faster.
The new index will cause that some old parts in YaCy can be removed,
i.e. specialized webgraph data and the noload crawler. The new index
will make it possible to:
- search within link texts of linked but not indexed documents (about 20
times of document index in size!!)
- get a very detailed link graph
- enhance ranking using a complete link graph
To get the full access to the new index, the API to solr has now two
access points: one with attribute core=collection1 for the default
search index and core=webgraph to the new webgraph search index. This is
also avaiable for p2p operation but client access is not yet
implemented.
structure, but is not filled yet. To have the opportunity of a second
core, multi-core functionality had to be implemented to the
deep-embedded solr:
- migrated the solr_40 directory content to a subdirectory
'collection1'; the previously used default core is now called
collection1
- added solr_40/webgraph subdirectory as second core
- added a servlet configuration for the second core 'webgraph' in
/IndexSchema_p.html
- added instance handling as addition to solr connections: all solr
connectors are now instances of an solr 'instance' object; this required
a complete re-design of the solr embedding
- migrated also caching and sharding ontop of new instance handling
- migrated the search apis to handle now the access to a specific core,
the default core named 'collection1'
- migrated the remote solr search interface to access shards of cores;
for the yacy remote search the default core is now called 'solr'; using
the peer address as solr address
- migrated the solr backup and restore process: old backups cannot be
used after this migration!
- redesign of solr instance handling in all methods which access the
instances: they cannot hold copies of these instances any more; the must
retrieve the actuall connection object every time they want to write to
it (this solves also some bugs when switching the index/network)
- added another schema 'solr.webgraph.schema', the old solr.keys.list is
replaced by solr.collection.schema
multiple solr cores instead of just one. Therefore it is now necessary
to distuingish between solr server connections (called an 'Instance')
and a connection to a single solr core. One Instance may now have
multiple connector classes assigned to it, each connecting to a single
core.
To support multiple cores it is also necessary to distinguish between
the connection configuration and the configuration of the index schema.
We will have multiple schema configurations in the future, each for
every solr core. This caused that the IndexFederated servlet had to be
split into two parts, the new Servlet for the Schema editor is now in
the IndexSchema Servlet.
one request:
- allow larger match-fields in html interface
- delete all host hashes at once from zurl
- when deleting by host, do not count size of deleted entries since that
was the reason it took so long
4.0.0 there is a new softcommit feature which implements a
near-real-time (NRT) search option. The softcommit does not do IO and
does not cause performance issues.
YaCy has now an extension in its solr connectors to use the softcommit
feature. The softcommit call now replaces all places where a hard commit
was used. Furthermore the commit strategy in when doing a search from
the web interface was changed (it's done every time before a search is
done).
The softcommit feature was implemented because it was needed for the
following changes (customer demands), which is also included in this
git commit:
- added a feature to identify all documents which have unique titles
and/or unique descriptions. These unique flags are disabled by default.
- added also a feature to set a flag when the url from a canonical tag
is equal to the document url. This is also disabled by default.
To support the new softcommit strategy, the commitWithinMs option was
set to -1 do disable automatic commit based on document insert times. If
documents are inserted permanently then also a commit would happen
permanently whenever the commitWithinMs time is reached. This would
conflict with the regular autocommit of 10 minutes and the new
softcommit strategy.
- migrates all entries in old urldb
Metadata coordinate (lat / lon) NumberFormatException still relative often (see excerpt below),
- added try/catch for URIMetadataRow (seems not to be needed in URIMetaDataNode, as Solr internally checks for number format)
- removed possible typ conversion for lat() / lon() comparison with 0.0f, changed to 0.0 (leaving it to the compiler/optimizer to choose number format)
current log excerpt for NumberFormatException:
W 2013/01/14 00:10:07 StackTrace For input string: "-"
java.lang.NumberFormatException: For input string: "-"
at sun.misc.FloatingDecimal.readJavaFormatString(Unknown Source)
at java.lang.Double.parseDouble(Unknown Source)
at net.yacy.kelondro.data.meta.URIMetadataRow$Components.lon(URIMetadataRow.java:525)
at net.yacy.kelondro.data.meta.URIMetadataRow.lon(URIMetadataRow.java:279)
at net.yacy.search.index.SolrConfiguration.metadata2solr(SolrConfiguration.java:277)
at net.yacy.search.index.Fulltext.putMetadata(Fulltext.java:329)
at transferURL.respond(transferURL.java:152)
...
Caused by: java.lang.NumberFormatException: For input string: "-"
at sun.misc.FloatingDecimal.readJavaFormatString(Unknown Source)
at java.lang.Double.parseDouble(Unknown Source)
at net.yacy.kelondro.data.meta.URIMetadataRow$Components.lon(URIMetadataRow.java:525)
at net.yacy.kelondro.data.meta.URIMetadataRow.lon(URIMetadataRow.java:279)
at net.yacy.search.index.SolrConfiguration.metadata2solr(SolrConfiguration.java:277)
at net.yacy.search.index.Fulltext.putMetadata(Fulltext.java:329)
at transferURL.respond(transferURL.java:152)
metadata and old rwi and for the citation index. The important
advancement is the separation of the citation index deletion because
that index is responsible for the linkdepth calculation. Now a search
index can be deleted without the citation index and that should cause
that less clickdepths must be post-processed.
This attribute can be used for ranking and for other purpose (demand by
customer)
The click depth is computed in two steps:
- during indexing the current fill-state of the reverse link index is
used to backtrack the current page to the root page. The length of that
backtrack is the clickdepth. But this does not discover the shortest
click depth. To get this, a second process to check again is needed
- added a process tag that can be used to do operations on the existing
index after a crawl; i.e. calculation the shortest clickpath. Added a
field to control this operation but not a method to operate on this.
- added a visualization of the clickpath length in the host browser
clicks which are necessary to get from the portal of a host to a
specific document. At this time, only the start document is flagged with
clickdepth '0', all other with '-1'. To get the actual clickdepth, a
process must use crawled information to collect the actual number of
clicks. This will be added in another/next step.
INCOMING links to the corresponding web page. This information is taken
from the reverse link index (a 'little sister' of the RWI index).
- this field can be of use to enhance the ranking because a web page
with more incoming links can be more more important than others. But
this is not true for typical link pages like menues. Therefore the
number of outgoing links is needed.
- added a new solr attribute 'bf' to solr queries which is a boost
function extension. this field can contain a formula which comuptes the
boost according to given field values. After some experiments the
following forumla is now default:
div(add(1,references_i),pow(add(1,inboundlinkscount_i),1.6))^0.4
This takes the number of references and the inbound links. Further
experiments are needed to enhance that forumula.
This uses an enhanced version of the Nutch/Solr TextProfileSignatue.
As a result, a signature of the document is written to the solr search
index. Additionally for each time when a signature is written, it is
checked if the singature exists already in the index. If the signature
does not exist, the document is marked as unique. The unique attribute
can now be used to sort document lists and bring duplicates to the end
of a result list.
To enable this, a large portion of the search api to Solr had to be
changed. This affected mainly caching of 'exists' searches to enhance
the check for existing signatures and do this without actually doing a
solr query.
Because here the first time a long number is used as value in the Solr
store, also the value naming in the YaCySchema had to be adopted and
normalized. This caused that many files had to be changed.
the actual data which is fetched from solr.
- used the new field options to reduce generic options like getting the
load date or the count of search results. should increase overall speed
- used the new field options to reduce overhead in the host browser
during aquisition of links.
- used the field options to make checking of links in crawler faster
- if the crawler is paused, the crawl queue is not cleaned
added solr highlighting / YaCy snippets to YaCy search results
- facets are now much more complete
- facets are computed and searched much faster
- snippet computation is done by solr if solr knows the snippet
- only visible for admins
- a faceted search generates a huge list for all hosts in the host list
- the faceted search algorithms had to be modified for that
- within the browsing of the directory path, the error cause is written
to the url which is presented as error-url
- the errors are also accumulated for directory sums
deleted which are selected by a crawl filter (host or subpath)
- site crawl used this option be default now
- made option to deleteDomain() concurrency
- time-out after 3 seconds to speed up display (may be incomplete)
- showing also all links from the balancer queue in the host list (after
the '/') and in the result browser view with tag 'loading'
superfluous. The target is to make a solr document as the core of YaCy
documents which would cause that many conversions can be removed. On the
way to this target the Equivalence of URIMetadataRow and URIMetadataNode
had to be removed to expose the usage of the old URIMetadataRow data
structure.
This refactoring already removes unneccessary conversions and should
make memory usage during indexing lower.
which had been then mixed with remote RWIs. Now these Solr documents are
feeded into the result set as they appear during local and remote
search. That makes the search much faster.
URIMetadataNode which creates the opportunity to access Solr objects
directly and use their information richness
- lazy initialization of the URIMetadataNode object - should cause less
computation and memory usage during search.
- removed dead code
- showing also outbound links to other domains if there are any
- the outbound links browser shows also the link structure image
- showing even inbound links if the web structure graph has information
about that
- removed the left menu and made the HostBrowser a part of the top menu
for search
- moved the file search also to the top menu
- added hover information in the HostBrowser to explain what the click
means
- because the HostBrowser also links to the Metadata viewer ViewFile,
there should be a button to switch back to the HostBrowser: added that
also.
MultiProtocolURI during normalform computation because that should
always be done and also be done during initialization of the
MultiProtocolURI Object. The new normalform method takes only one
argument which should be 'true' unless you know exactly what you are
doing.
there is no need for the RWI index if the index is not transfered to
another peer. Therefore the creation of RWI index data is now suppressed
if DHT is disabled. This applies for all intranet and portal mode
configurations, but not for public robinson modes. A robinson may switch
back to public mode and then transmit its data. That means if someone
wants to switch never to DHT mode, it would be more appropriate to
choose the portal mode.
This can be used to add another stemming to solr using stemming files
that are expressed as synonyms for grammatical alternatives. The
synonym/stemming files must have the following form:
- each line is a comma-separated list of synonyms
- the list of synonyms may be enclosed with {} (like the GSA synonyms
file)
- the file may contain comments which are lines starting with a '#'
The synonym file(s) must be placed in DATA/DICTIONARIES/synonyms/ and
are activated by default whenever a synonym file is in place.
Then, for each word that is found in a document all synonyms are added
to a long text field which is stored into synonyms_t. Processes using
the synonyms must query with that field as optional matcher.
Therefore a property like solrConnected() must be true all the time.
Removal of this method causes removal of all write operations to the old
metadata index.
create an document entry. This makes remote search much slower.
- removed synchronization of add method if ip_s is activated to prevent
that a user configuration causes bad behavior. The disadvantage of that
is, that a index dump can cause data loss if an indexing is running
during index dump
- catched more exceptions and more NPE
- better abstraction in MirrorSolrConnector
- slight performance enhancement when only the index count is requested
(rows=0 is sufficient to get a total count)
dump commands
- adjusted the apicall.sh script to get the downloaded text as output to
stdout which is necessary to parse the content out of it
- added indexdump.sh script which creates a solr dump and prints out the
storage path for the index dump
- added synchronization to the Fulltext class to prevent that data is
stored to a non-existing solr index while this index is disabled during
the storage of the dump
