[CWB] Where to start looking for the problem?

"Andrés Chandía" andres at chandia.net
Wed Apr 20 19:23:10 CEST 2022 


Thanks Jörg,


I've done what you suggested:


Delete cache overflow   
Clear entire cache (but keep saved queries)
Delete DB cache overflow
Clear entire DB cache



But things remain the same.


On the other hand, by shell things work:


[no corpus]> info COCA;
Warning:
    Can't open info file
/mnt/vmdata/corptedig-glif/corpora/cqp/data/coca/.info for reading
Size:    679142308
Charset: utf8
Properties:
        language = '??'
        charset = 'utf8'

[no corpus]>
COCA;
COCA> show cd;
===Context
Descriptor=======================================

left
context:     25 characters
right context:    25
characters
corpus position:  shown
target anchors:   not shown

Positional Attributes:  * word
                         
pos
                         
lemma

Structural Attributes:    text
                         
text_id              [A]

Aligned Corpora:          <none>

============================================================
COCA>
"king";
    89847: @!JAMES-CAMERON# I 'm the <king> of the
world . It 's jus
   158684: tar maker himself and the <king> of late
night , Johnny C
   202194: L-BROWN# My baby got me a <king> throne chair
. @!ELIZABE
   377023: L-BROWN# My baby got me a <king> throne chair .
@!ELIZABE
   407693: e the bacon . This is the <king> of Bacon talking
about b
   460068:  is allowed to say to the <king> , we ai n't
giving you m
   569597: ) The year was 1983 , the <king> of pop ruled the
music s
   758961: edding sets each queen or <king> . @!NATALIE-MORALES#
Yea
   872891: TALIE-MORALES# Who is the <king> of the pun . AL ROKER :

   947040: g the names of the future <king> or queen . @(Begin-VT) @
   978940: u 're the dessert mash-up <king> . This is fantastic . @!
  1069401:  he is quickly becoming a <king> of TV , books and radio 
  1103815: T# Look at this guy , the <king> of the segue , incredibl
 
1124892: power vested in me as the <king> of Rock ' N ' Roll , con
  1147397:
name Rory which means red <king> in Irish . And so , henc
  1201854: # But you
are the weather <king> . So go for it . AL ROKE
  1265790: e newest Hobbit
movie was <king> of the Box Office this w
  1365468: . @!MICHAEL-ALLER# He was
<king> of all the world ; " and
  1416028:  should be treated like a
<king> . And Chef Marco Moreira
  1522053: -DUBOIS# Japan is the new
<king> of the music hill , @ @ 
  1602009: n 1977 . America lost its
<king> . Do you remember who it
  1602038: six years ago today , the
<king> of rock and roll , Elvis
  1653407:  to the blues , he 's the
<king> of kings . There have be
  1724302: monarchy . They 've got a
<king> and a majority Shia nati
  1746450: rejected the concept of a
<king> when our country was fou
  1864103: s is in no rush to become
<king> . In an interview with C
  1864134: nce of Wales compares the
<king> 's role to a form of pri
  2213017: ed , Jordan ca n't -- the
<king> of Jordan ca n't last . 
  2379213: t north but they captured
<king> side ( ph ) aircraft . A
  2392055: t north but they captured
<king> side ( ph ) aircraft . A
  2513172: nd Senator Rob Byrd , the
<king> of the Senate at the tim
  2550030: hey can announce a future
<king> or an election win . The
  2698722: asically , especially the
<king> crab , about half goes t
  2835651: ss the board and see what
<king> of legislation they are 
  2837847: pt and we do n't have any
<king> to take over again . @!B
  3263564: sident ? Was it : ( a ) ,
<king> , ( b ) , the president 
  3302053: mehow , here 's queen and
<king> , President Obama and Jo
  3302078: ckmated the queen and the
<king> with the help of one lit
  3302094:  the bishop takes out the
<king> . Somehow that happened 
  3302116: playing with this kind of
<king> , maybe something like t
  3303735: e and put yourself as the
<king> of the Middle East , you
  3501709: at we 're going to have a
<king> again in America . So on
  3590794: w , here is this reigning
<king> whose life is really tak
  3609279: ved in a coup against the
<king> . The Scientologists wer
  3609291: at an attempt to kill the
<king> at a military parade and
  3666078: ot walk alone , to hear a
<king> proclaim that our indivi
  3742998: dinand and Isabella , the
<king> and queen we know from t
  3743499:  is real or not , but the
<king> and queen decided to sig
  3743551:  give us a pass . And the
<king> and queen thought about 
  3821267:  n't seem too slow when a
<king> or a mob come in and do 
  3896067: FORD# My mom is truly the
<king> of this game , though . 
  3938592: e @ @ @ @ @ @ @ @ @ @ the
<king> of late night . Ferguson
  3961624: ry funny man . He was the
<king> of variety television in
  3982726: you know , but he was the
<king> and you could n't ... @!
  4028273:  is supreme , the male is
<king> , no matter what the dif
  4208364: z around one-time cycling
<king> Lance Armstrong came to 
  4330533: @ @ @ @ @ @ @ @ @ @ . The
<king> was the first Arab leade





El Mie, 20 de Abril de 2022, 18:03, Jörg Knappen escribió:
 


A few things to check ...

> "The query ran successfully in 0 seconds."

So the results may be cached, clear the cache in the admin interface to preclude a cache
issue. I have experienced strange query failures from cached queries when the first query ran
out of time because of a temporary server overload.




Try the queries on the cqp command line: are the results OK than?




--Jörg Knappen



Am 2022-04-20 17:51, schrieb Andrés Chandía:

Hi Andrew,

Thanks for your help, I have changed
$print_debug_messages = true;

but all I see besides all that I have reported before
is: "The query ran successfully in 0 seconds."

"BTW While 3.2.27 is
old, it’s not that old relative to the latest in the branch"

Yes, I have
not been in charge of this server for a long time, actually the corpus I'm having the problem
with was not istalled by me, and the state of the server is as I have left it long time
ago.

At first I thought it could be an issue of unupdated things and I did the
updating of the SO, having some issues with "AppArmor+MySql" which I could solve,
but that didn't help in the corpus issue, by the way, I have checked the corpus I have
installed when I was in charge of this server and all of them work correctly...

So,
as I see it, I think the best solution is to install this corpus again, what do you think
about?
 
 
 
 <style type="text/css">-></style>


Hi Andrés, 
 
Wow, what a pile of issues!
 
The key point is this one: 
 
 There is a corpus that I've been told it
was working correctly, but all of a sudden queries started to give back [UNREADABLE]
 
This can’t happen without cause. Something
must have changed. The range of other error messages suggests that the setup of the operating
system has changed. Perhaps it has been upgraded, or default settings restored. In any case,
the issues would seem to be with configuration of either the filesystem, or the MySQL daemon,
or both.
 
 So, my question is where and how should
I start to try to find out the cause of this problem?
 
Turn on showing debug messages in the config
file. This should help. 
 
Note that “[UNREADABLE]” is caused
by failure to parse concordance output from CQP. So, seeing what is actually being passed back
and forth will probably give you a hint. 
 
BTW While 3.2.27 is old, it’s not
that old relative to the latest in the branch. 
 
best
 
Andrew.
 


From: cwb-bounces at sslmit.unibo.it On Behalf Of
"Andrés Chandía"
Sent: 20 April 2022
14:59
To: Open source development of the Corpus WorkBench 
Subject: [CWB] Where to start looking for the problem?


 


Hi there,


 


At a cqpweb installation (CQPweb v3.2.27)


 


There is a corpus that I've been told it was working correctly, but all
of a sudden queries started to give back [UNREADABLE]


 


Like:


 


[UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE]
[UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE][UNREADABLE] [UNREADABLE] [UNREADABLE]
[UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE]
[UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE] [UNREADABLE]
[UNREADABLE] [UNREADABLE]


 


So, my question is where and how should I start to try to find out the
cause of this problem?


 


Things I've done/notice:


 


If I click on the linked results and the switch to the alternative view
(pos), I get:


 


i ge mc fo ) rr , rr21 rr22 , pn1 vvd np1 np1 , cc np1 np1 vvd ii21
ii22 nn1 ii21 ii22 nnt1 cs pphs1 vdd xx vvi to vbi rl to vvi ppge at _appge - appge nn1 vvg dd1
nn1 . 

rr pph1 vbdz xx rr at1 nn1 nnt1 . 

ccb md - md_nnt1 nn2 vh0 vbn -
vh0 vbn jj . 

cs ppis1 vbdr np1 , ppis1 vm xx vbi vvg_jj@ nn1 ii dd1 io dd2 nn2 .



 


If I go to the Admin Control Panel I see:


 


Corpus: coca | Indexing date: 2019-04-11 15:29:37 | Size Tokens: 0 |
Types: 1,274,893 | Texts: 0 | Disk space Indexes: 0.0 MB | Freq tables: 57.4 MB


 


Frequency list search show results correctly:


1 king 11,825


 


and all derived from king...


 


A word lookup will fail giving this output: A MySQL query did not run
successfully!



Original query: select count(concat(node,'_',tagnode)) as tokens,
count(distinct(concat(node,'_',tagnode))) as types from db_sort_h1fel9jyg4 /* from User: admin
| Function: require() | 2022-Apr-20 13:37:27 */

Error # 1054: Unknown column 'tagnode' in 'field list'


 

Generate CWB text-position records outputs: A MySQL query did not run
successfully!
Original query: insert into `___temp_cqp_text_positions_for_coca` (text_id,
cqp_begin, cqp_end) VALUES ('4122770', 0, 5307), .... etc. a lot of numbers...
Error #
1062: Duplicate entry '4166449' for key 'PRIMARY'

 


Update Word and file counts outputs "the connections has
expired", on retrying page keeps loading until is done


 


Recreate CWB Frequency table outputs: "the connections has
expired" on retrying I get: CQPweb could not create a directory for the frequency index.
Check filesystem permissions!


 


I've checked the directories and all of them are owned by www-data


 


 


Recreate Frequency tables outpus: "the connections has
expired" on retry finally outpus;


 



A MySQL query did not run successfully!

Original query: CREATE TABLE __tempfreq_coca ( freq int(11) unsigned default NULL, word
varchar(255) NOT NULL, lemma varchar(255) NOT NULL, pos varchar(255) NOT NULL, key (word), key
(lemma), key (pos) ) CHARACTER SET utf8 COLLATE utf8_general_ci /* from User: admin |
Function: corpus_make_freqtables() | 2022-Apr-20 13:53:57 */

Error # 1050: Table '__tempfreq_coca' already exists


 


I will appreciate any guidance, thanks a lot!!


 


 


 

_______________________
andrés chandía

Düngupeyem |  IECMap |  ISECMap |  NMT |  Corlexim

Desarrollador de:
Parles.upf |  IWCH |  Amind terapia |  Nocando |  IAC |  CddZ |  ISAC |  CatCg
P  No imprima innecesariamente. ¡Cuide el medio ambiente!






_______________________
            andrés chandía
 
Düngupeyem | IECMap
| ISECMap | NMT | Corlexim

Desarrollador de:
Parles.upf | IWCH | Amind terapia | Nocando | IAC | CddZ | ISAC | CatCg
P No imprima innecesariamente. ¡Cuide el medio
ambiente!


_______________________________________________
CWB mailing list
CWB at sslmit.unibo.it
http://liste.sslmit.unibo.it/mailman/listinfo/cwb





 


_______________________

            andrés
chandía
 
Düngupeyem | IECMap | ISECMap | NMT | Corlexim

Desarrollador de:
Parles.upf | IWCH | Amind terapia | Nocando | IAC | CddZ | ISAC | CatCg
P No imprima innecesariamente. ¡Cuide el
medio ambiente!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://liste.sslmit.unibo.it/pipermail/cwb/attachments/20220420/ed585d60/attachment-0001.html>

More information about the CWB mailing list