Hi Andi,
thanks a lot for your support - it works!
Unfortunately I have more than one TDB store. So I created a small script to
fix this issue. I agree that it is ugly, but it worked for me.
#!/bin/bash
cd apache-jena-fuseki-2.0.0/run/system_files/
for f in *
do
cat $f | sed 's/3000/3000000/g' > ../configuration/$f.ttl
done
cd ..
rm -r system
And thanks for addressing this issue in JIRA!
Cheers
Sebastian
-----Ursprüngliche Nachricht-----
Von:
users-return-11152-sebastian.mate=imi.med.uni-erlangen...@jena.apache.org
[mailto:users-return-11152-sebastian.mate=imi.med.uni-erlangen.de@jena.apach
e.org] Im Auftrag von Andy Seaborne
Gesendet: Sonntag, 19. April 2015 18:36
An: [email protected]
Betreff: Re: Fuseki 2 ignoring timeout setting
Hi Sebastian,
Found it - the default configuration file for TDB created by the UI has
a 3s default query timeout; it shouldn't. Sorry about that.
Workaround:
The directory run/system_files has a copy of the file. This is not the file
the server works off (that's in the database).
0/ Stop the server
1/ If you have only one dataset, copy that file to
run/configuration/mydatabase.ttl (any fresh base name, must end ttl).
2/ Edit the "run/configuration" file to remove the line:
ja:context [ ja:cxtName "arq:queryTimeout" ; ja:cxtValue "3000" ] ;
or set it to whatever you want.
3/ Delete the system database
4/ Restart the server
You can dump the system database (tdbdump; when the server is stopped) and
look to see if there is anything there. It's a regular TDB database, albeit
with custom setup to make it small (reduced caches, block size is 1k, it's
direct mode only).
I agree that keeping the persistent state in a TDB database makes its
opaque. That is something worth reconsidering. It could be all file based
which makes hacking it easier.
Andy
Record as JENA-918
On 17/04/15 19:52, Sebastian Mate wrote:
> Hi Andy,
>
> thanks for your quick reply!
>
> No, I did not configure timeouts anywhere else. I created the TDB
> datasets using the Fuseki web interface and then used tdbloader2 to
> upload the triples into their respective TDB stores (with a Bash
> script to make life easier).
>
> Where else can I set this timeout option? Do I have to modify
> run/config.ttl?
>
> Just guessing - correct me if Im wrong: It seems to me that the TDB
> store itself is configured in RFD. Duh - this makes life much more
> complicated! Where can I find more information about this? Is this
> specific to Fuseki or TDB?
>
> Sebastian
>
>
> Am 17.04.2015 um 19:48 schrieb Andy Seaborne <[email protected]>:
>
>> Hi Sebastian,
>>
>> Are timeout also set anywhere else such as the service/dataset
configuration? If so, they will override the global (server-wide) default.
>>
>> The 503 message is confusing - it prints the gloabl default (100000ms)
even if the query executed under a different timeout. It looks like it it
set to 3000ms somewhere else as well.
>>
>> Andy
>>
>> On 17/04/15 13:49, Mate, Sebastian wrote:
>>> Dear Fuseki users,
>>>
>>> I want to use Fuseki 2 for querying various big ontologies, e.g.
dbpedia. However, I have encountered the problem that Fuseki is ignoring the
timeout setting.
>>>
>>> I started the server with
>>>
>>> ./fuseki-server --timeout=100000
>>>
>>> which should set the timeout to 100s, but it is rather 3s (see below). I
have seen that there were some discussions regarding the handling of
timeouts, but I'm not sure if this is related to my problem. Here's the log
- please note that I'm using one of the latest snapshot builds (the official
2.0.0 release is behaving the same way for me):
>>>
>>> [2015-04-17 09:53:14] Server INFO Fuseki 2.0.1-SNAPSHOT
2015-04-16T16:45:46+0000
>>> ...
>>> [2015-04-17 09:53:58] Config INFO Register:
/dbpedia_topical_concepts_unredirected_en
>>> [2015-04-17 09:53:58] Config INFO Register:
/dbpedia_wikipedia_links_en
>>> [2015-04-17 09:53:58] Server INFO Started 2015/04/17 09:53:58 MESZ
on port 3030
>>> [2015-04-17 09:54:12] Fuseki INFO [1] POST
http://xxx.xxx.xxx.xxx:3030/dbpedia_page_links_unredirected_en/query
>>> [2015-04-17 09:54:12] Fuseki INFO [1] POST
/dbpedia_page_links_unredirected_en :: 'query' ::
[application/x-www-form-urlencoded charset=UTF-8] ?
>>> [2015-04-17 09:54:12] Fuseki INFO [1] Query = select (count(*) as
?count) {?s ?p ?o}
>>> [2015-04-17 09:54:15] Fuseki INFO [1] 503 The query timed out
(restricted to 100000 ms) (3,098 s)
>>>
>>> The problem appears with other queries as well. Sometimes limiting the
result set (e.g. by appending "LIMIT 20" to the SPARQL query) helps.
>>>
>>> The system is a quad core virtual machine with 32 GB RAM, running Ubuntu
14.04. "java -version" is reporting:
>>>
>>> OpenJDK Runtime Environment (IcedTea 2.5.4) (7u75-2.5.4-1~trusty1)
>>> OpenJDK 64-Bit Server VM (build 24.75-b04, mixed mode)
>>>
>>> I would highly appreciate any advice. I really like Fuseki and would
love to use it for some data mining in medical informatics.
>>>
>>> Many thanks,
>>> Sebastian
>>>
>>> --------------------------------------------------------------------
>>> ----------------------
>>> Dipl.-Inf. Sebastian Mate
>>>
>>> Lehrstuhl für Medizinische Informatik
>>> Friedrich-Alexander-Universität Erlangen-Nürnberg Wetterkreuz 13,
>>> 91058 Erlangen-Tennenlohe, Germany
>>>
>>> Tel. +49 (9131) 85-26722 // Fax +49 (9131) 85-26754
>>> eMail: [email protected]
>>>
>>>
>>>
>>>
>>>
>>
>