To conclude this thread, the discussion was restarted on sr-dev mailing list and a reworked patch is now in master branch.

Cheers,
Daniel

On 09/12/13 13:23, jay binks wrote:

Ok so... yea I should have waited another 10 min before posting to the list..

I did the patch... it works for me.

if someone familiar with db_cassandra can clean this up and commit it, I would be quite grateful.

# git diff dbcassa_base.cpp

diff --git a/modules/db_cassandra/dbcassa_base.cpp b/modules/db_cassandra/dbcassa_base.cpp

index e9d3a32..155221d 100644

--- a/modules/db_cassandra/dbcassa_base.cpp

+++ b/modules/db_cassandra/dbcassa_base.cpp

@@ -439,6 +439,7 @@ ColumnVecPtr cassa_translate_query(const db1_con_t* _h, const db_key_t* _k,

int key_len=0, seckey_len = 0;

int no_kc, no_sec_kc;

dbcassa_table_p tbc;

+ char pk[255];

/** Lock table schema and construct primary and secondary key **/

if(_k) {

@@ -495,8 +496,12 @@ ColumnVecPtr cassa_translate_query(const db1_con_t* _h, const db_key_t* _k,

} else { /* the table doesn't have any secondary key defined */

if(_c) {

for(int i=0; i< _nc; i++) {

- sp.column_names.push_back(_c[i]->s);

- LM_DBG("Query col: %s\n", _c[i]->s);

+ sprintf(pk, "%.*s", _c[i]->len, _c[i]->s );

+ sp.column_names.push_back( pk);

+ //sp.column_names.push_back(_c[i]->s);

+ LM_DBG("Query col: %s\n", pk );

+ //LM_DBG("Query col: %s\n", _c[i]->s);

+ LM_DBG("JAY Query col: %.*s\n", _c[i]->len, _c[i]->s);

}

LM_DBG("get %d columns\n", _nc);

sp.__isset.column_names = true; // set

yea I know I left crap in there, but it give syou an idea... also...

yea pk is a crap name... copy paste... and I havnt given any thought to the size 255 , but thats probably fairly safe and anything smaller might not be so safe.

( unless there is a limit elsewhere I should observe )

Jay
On 9 December 2013 22:14, jay binks <jaybinks@gmail.com> wrote:
Ok my latest makes me think this is an issue in db_cassandra.

when I add the following debug line it displays as expected.

496 if(_c) {

497 for(int i=0; i< _nc; i++) {

498 sp.column_names.push_back(_c[i]->s);

499 LM_DBG("Query col: %s\n", _c[i]->s);

500 LM_DBG("JAY Query col: %.*s\n", _c[i]->len, _c[i]->s);

501 }

502 LM_DBG("get %d columns\n", _nc);

503 sp.__isset.column_names = true; // set

so _c[i]->s is just going to the end of the string, instead of respecting _c[i]->len

While I COULD work out a fix for this using sprintf or similar, it would probably not be in keeping with kamaiio convention.
( since im not experienced with kamailio code )
Also, im not sure what other places I would miss.

can someone more experienced with kamailio take a look and propose a patch for me to test.

Thanks

Jay
On 9 December 2013 21:34, jay binks <jaybinks@gmail.com> wrote:
kamailio -v

version: kamailio 4.2.0

im still digging, but my current hunch is that my prior diagnosis may be wrong, and this might be in db_cassandra ...

still digging, but I will update the list with my findings if they are of interest.
On 9 December 2013 21:26, Daniel-Constantin Mierla <miconda@gmail.com> wrote:
What version are you using?

Daniel

On 09/12/13 12:17, jay binks wrote:
So when Im trying to use is_subscriber & load_credentials together, It seems I have either hit a bug or just a bad config ( but I dont think its the latter )

My symptom is that only the field on the far right of "load_credentials" ever returns the correct value, the other fields are always null.

My config has this setting :

modparam("auth_db", "load_credentials", "field1;field2")

and I can see the following debug when running kamailio -ddddd

0(14174) DEBUG: auth_db [authdb_mod.c:291]: w_is_subscriber(): uri [sip: 1112@192.168.173.5] table [subscriber] flags [1]

0(14174) DEBUG: db_cassandra [dbcassa_base.cpp:807]: db_cassa_query(): query table=subscriber

0(14174) DEBUG: db_cassandra [dbcassa_table.c:461]: dbcassa_db_get_table(): found dbname=kamailio, table=subscriber

0(14174) DEBUG: db_cassandra [dbcassa_base.cpp:351]: cassa_constr_key(): query col = username

0(14174) DEBUG: db_cassandra [dbcassa_base.cpp:351]: cassa_constr_key(): query col = domain

0(14174) DEBUG: db_cassandra [dbcassa_base.cpp:376]: cassa_constr_key(): key = 1112 192.168.173.5

0(14174) DEBUG: db_cassandra [dbcassa_base.cpp:499]: cassa_translate_query(): Query col: field2

0(14174) DEBUG: db_cassandra [dbcassa_base.cpp:499]: cassa_translate_query(): Query col: field1; field2

0(14174) DEBUG: db_cassandra [dbcassa_base.cpp:501]: cassa_translate_query(): get 2 columns

0(14174) DEBUG: <core> [db_res.c:118]: db_new_result(): allocate 56 bytes for result set at 0x7fc6ce770e78

if I had to put money on where the bug is, I would look in parse_aaa_pvs in authdb_mod.c

however, an initial glance revealed nothing glaring, also this file was last changed in 2013-01-20 ( and that was only a move from modules_k )

SURELY I cant be the only person wanting to use this for more than one field ?

This code looks like its almost the same as when first committed in August 2011 ....

Id love someone to see if they can confirm this and point me in the right direction ( if this is my fault ) or lets get this fixed, if its a bug :)

--
Sincerely

Jay
_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
-- Daniel-Constantin Mierla - http://www.asipto.com http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda
_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
--
Sincerely

Jay
--
Sincerely

Jay
--
Sincerely

Jay

-- 
Daniel-Constantin Mierla - http://www.asipto.com
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda