date:20200113

Re: [Maria-discuss] [Maria-developers] [MariaDB 10.3.7] Hello, everyone. My process core dumped while calling mysql_ping().

2020-01-13 Thread Markus Mäkelä


Hi,


Is the connection protected by a lock of some sorts? To my knowledge the 
library does not support concurrent use of a single connection. I'd 
recommend trying to first wrap all access to the connection with a mutex 
and see if that solves the problem.



Markus


On 1/14/20 08:33, Ma Allen wrote:

Environment:
MariaDB 10.3.7 database server & related headers and libraries
CentOS 7.3 or 7.4

./configure --prefix=/home/spiderflow --localstatedir=/home/spiderflow 
--enable-unix-socket --with-libnss-libraries=/usr/lib64 
--with-libnss-includes=/usr/include/nss3 
--with-libnspr-libraries=/usr/lib64 
--with-libnspr-includes=/usr/include/nspr4 --enable-netmap 
--enable-rust --disable-gccmarch-native --enable-mysql 
--with-libmysql-includes=/usr/local/mariadb-10.3.7-linux-systemd-x86_64/include/mysql 
--with-libmysql-libraries=/usr/local/mariadb-10.3.7-linux-systemd-x86_64/lib


Phenomenon:
(gdb)
#0  ma_pvio_write (pvio=0x10101010101b2e0, 
buffer=buffer@entry=0x2510101 "\001", length=length@entry=5)

    at /home/buildbot/buildbot/build/libmariadb/libmariadb/ma_pvio.c:350
#1  0x7f7100249dea in ma_net_real_write (net=net@entry=0xee9de0 
, packet=0x2510101 "\001", len=5)

    at /home/buildbot/buildbot/build/libmariadb/libmariadb/ma_net.c:335
#2  0x7f7100249f60 in ma_net_flush (net=net@entry=0xee9de0 
) at 
/home/buildbot/buildbot/build/libmariadb/libmariadb/ma_net.c:166
#3  0x7f710024a71a in ma_net_write_command (net=net@entry=0xee9de0 
, command=command@entry=14 '\016', 
packet=packet@entry=0x7f71003d6c67 "", len=0, 
    disable_flush=disable_flush@entry=0 '\000') at 
/home/buildbot/buildbot/build/libmariadb/libmariadb/ma_net.c:244
#4  0x7f7100252ca4 in mthd_my_send_cmd (mysql=0xee9de0 
, command=, arg=0x7f71003d6c67 "", 
length=0, skipp_check=, 
    opt_arg=) at 
/home/buildbot/buildbot/build/libmariadb/libmariadb/mariadb_lib.c:394
#5  0x7f71002518d0 in mysql_ping (mysql=0xee9de0 ) 
at /home/buildbot/buildbot/build/libmariadb/libmariadb/mariadb_lib.c:2552
#6  0x006651b2 in MySqlWaitConnected 
(p_mysql_data=p_mysql_data@entry=0xee9de0 ) at 
../srcSF/spiderFlow-proto-baseline.c:676
#7  0x00668819 in LoadProtoConfInfos 
(pInfo=pInfo@entry=0xee9de0 ) at 
../srcSF/spiderFlow-proto-baseline.c:835
#8  0x0066c53a in InitBaseLine () at 
../srcSF/spiderFlow-proto-baseline.c:1069
#9  0x005f308b in PostConfLoadedSetup (suri=0x10dbb80 
) at suricata.c:2900
#10 0x00425ce8 in main (argc=5, argv=0x7fff1cdca258) at 
suricata.c:3072

(gdb)

Analyzation:
The core dump always exists after running for a while or maybe after 
the sever restarts. It occurs at the initialization and involves with 
ONLY ONE thread, which is the main thread of the whole process.
The related code is two functions, which both checks the connection 
and read data from MariaDB. The first function works fine, however, 
the second one's checking connection leads to core dump.


The code to check connection is as follows:
void MySqlWaitConnected(MYSQL * p_mysql_data)
{
while (mysql_ping(p_mysql_data) != 0) {
SCLogInfo("Mysql ping error:%s", mysql_error(p_mysql_data));
#ifdef WIN32
Sleep(3000);
#else
sleep(3);
#endif
}
}

Previously, I solved a likely similar problem as follows:
Multiple threads shared ONE MySQL handler, which is the MYSQL data 
structure. Before inserting data into MariaDB without mutex, 
MySqlWaitConnected() is called. So while multiple threads do the same 
operation, inserting data is interrupted by mysql_ping() that leads to 
core dump. Afterwards, I comment MySqlWaitConnected() and add mutex 
for mysql_real_query() and mysql_commit().


Does anyone meet the same problem before? Any suggestion is 
appreciated. Thanks in advance.


Best Regards,
Allen Ma

___
Mailing list: https://launchpad.net/~maria-developers
Post to : maria-develop...@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-developers
More help   : https://help.launchpad.net/ListHelp


--
Markus Mäkelä, Senior Software Engineer
MariaDB Corporation
t: +358 40 7740484 | IRC: markusjm@freenode

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

[Maria-discuss] [MariaDB 10.3.7] Hello, everyone. My process core dumped while calling mysql_ping().

2020-01-13 Thread Ma Allen

Environment:
MariaDB 10.3.7 database server & related headers and libraries
CentOS 7.3 or 7.4

./configure --prefix=/home/spiderflow --localstatedir=/home/spiderflow 
--enable-unix-socket --with-libnss-libraries=/usr/lib64 
--with-libnss-includes=/usr/include/nss3 --with-libnspr-libraries=/usr/lib64 
--with-libnspr-includes=/usr/include/nspr4 --enable-netmap --enable-rust 
--disable-gccmarch-native --enable-mysql 
--with-libmysql-includes=/usr/local/mariadb-10.3.7-linux-systemd-x86_64/include/mysql
 --with-libmysql-libraries=/usr/local/mariadb-10.3.7-linux-systemd-x86_64/lib

Phenomenon:
(gdb)
#0  ma_pvio_write (pvio=0x10101010101b2e0, buffer=buffer@entry=0x2510101 
"\001", length=length@entry=5)
at /home/buildbot/buildbot/build/libmariadb/libmariadb/ma_pvio.c:350
#1  0x7f7100249dea in ma_net_real_write (net=net@entry=0xee9de0 
, packet=0x2510101 "\001", len=5)
at /home/buildbot/buildbot/build/libmariadb/libmariadb/ma_net.c:335
#2  0x7f7100249f60 in ma_net_flush (net=net@entry=0xee9de0 ) 
at /home/buildbot/buildbot/build/libmariadb/libmariadb/ma_net.c:166
#3  0x7f710024a71a in ma_net_write_command (net=net@entry=0xee9de0 
, command=command@entry=14 '\016', 
packet=packet@entry=0x7f71003d6c67 "", len=0, 
disable_flush=disable_flush@entry=0 '\000') at 
/home/buildbot/buildbot/build/libmariadb/libmariadb/ma_net.c:244
#4  0x7f7100252ca4 in mthd_my_send_cmd (mysql=0xee9de0 , 
command=, arg=0x7f71003d6c67 "", length=0, 
skipp_check=, 
opt_arg=) at 
/home/buildbot/buildbot/build/libmariadb/libmariadb/mariadb_lib.c:394
#5  0x7f71002518d0 in mysql_ping (mysql=0xee9de0 ) at 
/home/buildbot/buildbot/build/libmariadb/libmariadb/mariadb_lib.c:2552
#6  0x006651b2 in MySqlWaitConnected 
(p_mysql_data=p_mysql_data@entry=0xee9de0 ) at 
../srcSF/spiderFlow-proto-baseline.c:676
#7  0x00668819 in LoadProtoConfInfos (pInfo=pInfo@entry=0xee9de0 
) at ../srcSF/spiderFlow-proto-baseline.c:835
#8  0x0066c53a in InitBaseLine () at 
../srcSF/spiderFlow-proto-baseline.c:1069
#9  0x005f308b in PostConfLoadedSetup (suri=0x10dbb80 ) at 
suricata.c:2900
#10 0x00425ce8 in main (argc=5, argv=0x7fff1cdca258) at suricata.c:3072
(gdb)

Analyzation:
The core dump always exists after running for a while or maybe after the sever 
restarts. It occurs at the initialization and involves with ONLY ONE thread, 
which is the main thread of the whole process.
The related code is two functions, which both checks the connection and read 
data from MariaDB. The first function works fine, however, the second one's 
checking connection leads to core dump.

The code to check connection is as follows:
void MySqlWaitConnected(MYSQL * p_mysql_data)
{
while (mysql_ping(p_mysql_data) != 0) {
SCLogInfo("Mysql ping error:%s", mysql_error(p_mysql_data));
#ifdef WIN32
Sleep(3000);
#else
sleep(3);
#endif
}
}

Previously, I solved a likely similar problem as follows:
Multiple threads shared ONE MySQL handler, which is the MYSQL data structure. 
Before inserting data into MariaDB without mutex, MySqlWaitConnected() is 
called. So while multiple threads do the same operation, inserting data is 
interrupted by mysql_ping() that leads to core dump. Afterwards, I comment 
MySqlWaitConnected() and add mutex for mysql_real_query() and mysql_commit().

Does anyone meet the same problem before? Any suggestion is appreciated. Thanks 
in advance.

Best Regards,
Allen Ma
___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Windows C connector 3.1.6

2020-01-13 Thread Kenneth Penza

Good morning,

I have checked further and it looks that the package is not available as it
is not compiling as per (
https://buildbot.askmonty.org/buildbot/builders/win-connector_c_3).

Regards
Kenneth

On Mon, Dec 23, 2019 at 11:38 AM Kenneth Penza  wrote:

> Hi,
>
> I am trying to download the Windows C connector 3.1.6 build to test an SSL
> issue, however the Windows packages are not available from
> Connector-c-3.1.6
> .
>
> On checking the Connector-c-3.1.5 release, the required files are
> available. Will the 3.16 Windows build be available ?
>
> Thanks
> Kenneth
>
___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] [Maria-developers] [MariaDB 10.3.7] Hello, everyone. My process core dumped while calling mysql_ping().

[Maria-discuss] [MariaDB 10.3.7] Hello, everyone. My process core dumped while calling mysql_ping().

Re: [Maria-discuss] Windows C connector 3.1.6

3 matches

Site Navigation

Mail list logo

Footer information