[Info-Ingres] cursors SIGSEGV on close

DATABASE

    Sponsored Links

    Next

[Info-Ingres] cursors SIGSEGV on close

Postby martin.bowes » Wed, 28 Feb 2007 01:04:42 GMT

Hi Everyone,

I have a program which uses an 'outer' cursor to read through a table, 
and on the basis of that data controls several 'inner' cursors to read 
other data in several other tables. The inner cursors are opened and 
closed as they are required. Each is a dynamic cursor with several 
parameter place holders so each has its own iisqlda into which the 
query is prepared in advance, with the cursor open providing the 
parameters required

Has anyone got any idea why it is that if it reads a certain number of 
cursors that the final close of the outer cursor will cause a SIGSEGV?

gdb on the core dump shows:
(gdb) backtrace
#0  0x40413d32 in __libc_free (mem=0x80ee8f8) at malloc.c:3043
#1  0x402ccce9 in MEfree () at eval.c:41
#2  0x401f52a3 in IIcs_FreeState () at eval.c:41
#3  0x401f5052 in IIcsClose () at eval.c:41
#4  0x08049e61 in main () at eval.c:41
#5  0x403b0177 in __libc_start_main (main=0x8048f10 <main>, 
argc=2, ubp_av=0xbffff87c, 
    init=0x8048b54 <_init>, fini=0x804aa90 <_fini>, rtld_fini=0x4000e184 
<_dl_fini>, 
    stack_end=0xbffff86c) at ../sysdeps/generic/libc-start.c:129

This program is an ESQL/C clone of another program that executes in 
JDBC. I suspect - but as yet cannot prove - that JDBC program of 
occasionally crashing ingres with a SIGSEGV in the QEF.

Martin Bowes

Re: [Info-Ingres] cursors SIGSEGV on close

Postby martin.bowes » Thu, 01 Mar 2007 19:02:27 GMT

Hi Everyone,
A quick update...This SIGSEGV seems to be a problem on II2.6 only. 
The program executed perfectly well on IIR3 and Ingres2006.

Marty
--
Random Farscape Quote #2:
Aeryn - Who knows, we might end up making things better.
John  - With our record, do you think thats likely?



<?xml  version="1.0" ?><html>
<head>
<title></title>
</head>
<body>
<p><font face="Arial"><span style="font-size:10pt">Hi Everyone,</span></font></p>
<p><font face="Arial"><span style="font-size:10pt">A quick update...This SIGSEGV seems to be a problem on II2.6 only. 
The program executed perfectly well on IIR3 and Ingres2006.<br/>
<br/>
</span></font></p>
<p><font face="Arial"><span style="font-size:10pt">Marty</span></font></p>
<p><font face="Arial"><span style="font-size:10pt">--</span></font></p>
<div align="left"><font face="Arial"><span style="font-size:10pt">Random Farscape Quote #2:</span></font></div>
<div align="left"><font face="Arial"><span style="font-size:10pt">Aeryn - Who knows, we might end up making things better.</span></font></div>
<div align="left"><font face="Arial"><span style="font-size:10pt">John  - With our record, do you think thats likely?</span></font></div>
<div align="left"><br/>
</div>
<div align="left"></div>
</body>
</html>


Similar Threads:

1.[Info-Ingres] Ingres server crashed with SIGBUS error (and SIGSEGV)

At 1:58 PM -0700 7/23/07,  XXXX@XXXXX.COM  wrote:
>After upgrading from Ingres 2.6/0305 to Ingres 2.6/0604 SP5 for
>Solaris 9 (after Jul 3 2007)
>Ingres started crashing with SIGBUS and once with SIGBUS & SIGSEGV
>errors on DMF.
>
>...                    )'
>!(sc0e.c, 525) ERsend--->'YODA    ::[38598             ,
>000000010092cec0]:     User:  <Replicator Queue Management>

If the session named is the replicator queue management thread
most of the time, try fooling around with the rep_txq_size parameter
to see if you can make the errors go away or reduce their frequency.
(I assume you're using replicator.)

Was there any stack traceback in the errlog.log?

Do you have II_DBMS_LOG set?  (ingprenv)  If not, try
mkdir $II_SYSTEM/ingres/files/dbmslogs
ingsetenv II_DBMS_LOG $II_SYSTEM/ingres/files/dbmslogs/iidbms_%p.log

and see if you get anything useful in the dbms log the next
time it crashes.  (Note: the value of II_DBMS_LOG is just a
full pathname that has to be writable by user ingres.
It can be anywhere convenient.  The _%p appends the process
ID so that you don't overwrite the same file over and over.)

I don't know what this error is, but from the sound of it,
I think you'll need a patch to really fix it.

Karl

2.[Info-ingres] Tracking a SIGSEGV

Hi Everyone,

I'm getting a few SIGSEGV's (without stack dumps) on an II2.6/0306 
p11670 (int.lnx/00) installation.

The SIGSEGV's are mainly in the OPF but I also get a few in the QEF.

The ones in the QEF tend to crash the server.

I've managed to associate the crashes with some queries, but these 
queries are executed very frequently without problem and they are 
trivial select and insert queries. I've examined the tech base and all the 
SIGSEGV's listed are on much more complicated queries or against 
distributed databases - neither of which are in play here.

As an example, the insert query which caused the most recent QEF 
SIGSEGV and server crash was:
INSERT INTO job_error_log (job_name,error_msg,exception_dump) 
VALUES (?,?,?)
The only interesting thing about that table is that the exeception_dump 
is a long byte. It also has a fourth field (a date) with a default value. But 
as I say, this query executes perfectly well most of the time.

I have II_DBMS_LOG defined for the installation but the log file NEVER 
shows details about the crash. In fact it never shows any details beyond 
a few minutes after start-up.

I've tried all sorts of things to load the server, execute some queries in 
a manner that attempts to duplicate conditions in the server before the 
problems - and with no success. Quite frankly, I'm stumped on how to 
proceed from here. 

Anyone got any suggestions?
Anyone got any idea why my II_DBMS_LOG is so stumpy?

Martin Bowes
--
Random Duckman Quote #114:
King Chicken: How dare you insult me in front of my wife, whose still
dangerously coherent.


<?xml  version="1.0" ?><html>
<head>
<title></title>
</head>
<body>
<div align="left"><font face="Arial"><span style="font-size:10pt">Hi Everyone,</span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">I'm getting a few SIGSEGV's (without stack dumps) on an II2.6/0306 
p11670 (int.lnx/00) installation.</span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">The SIGSEGV's are mainly in the OPF but I also get a few in the QEF.</span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">The ones in the QEF tend to crash the server.</span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">I've managed to associate the crashes with some queries, but these 
queries are executed very frequently without problem and they are 
trivial select and insert queries. I've examined the tech base and all the 
SIGSEGV's listed are on much more complicated queries or against 
distributed databases - neither of which are in play here.</span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">As an example, the insert query which caused the most recent QEF 
SIGSEGV and server crash was:</span></font></div>
<div align="left"><font face="Times New Roman" size="3"><span style="font-size:12pt">I</span></font><font face="Arial"><span style="font-size:10pt">NSERT INTO job_error_log 
(job_name,error_msg,exception_dump) 
VALUES (?,?,?)</span></font></div>
<div align="left"><font face="Arial"><span style="font-size:10pt">The only interesting thing about that table is that the exeception_dump 
is a long byte. It also has a fourth field (a date) with a default value. But 
as I say, this query executes perfectly well most of the time.</span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">I have II_DBMS_LOG defined for the installation but the log file NEVER 
shows details about the crash. In fact it never shows any details beyond 
a few minutes after start-up.</span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">I've tried all sorts of things to load the server, execute some queries in 
a manner that attempts to duplicate conditions in the server before the 
problems - and with no success. Quite frankly, I'm stumped on how to 
proceed from here. </span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">Anyone got any suggestions?</span></font></div>
<div align="left"><font face="Arial"><span style="font-size:10pt">Anyone got any idea why my II_DBMS_LOG is so stumpy?</span></font></div>
<div align="left"><br/>
</div>
<div align="left"><font face="Arial"><span style="font-size:10pt">Martin Bowes</span></font></div>
<div align="left"><font face="Arial"><span style="font-size:10pt">--</span></font></div>
<div align="left"><font face="Arial"><span style="font-size:10pt">Random Duckman Quote #114:</span></font></div>
<div align="left"><font face="Arial"><span style="font-size:10pt">King Chicken: How dare you insult me in front of my wife, whose still</span></font></div>
<div align="left"><font face="Arial"><span style="font-size:10pt">dangerously coherent.</span></font></div>
<div align="left"></div>
</body>
</html>

3.[Info-Ingres] SIGSEGV from libscf.1.so(scs_iformat+0x5ff)

4.[Info-ingres] Iidbms crashing SIGBUS/SIGSEGV ; E_DM002B

5.[Info-ingres] Sigsegv on optimize of imadb

Hi Everyone,

I've just noticed that if I run optimizedb -zk -u'$ingres' imadb on IngresII 
R3 or Ingres2006 I get a SIGSEGV at the front-end (and in the errlog).
At the front-end only I get: 
E_OP095B        System or hardware exception occurred in optimizer 
utility.

The optimizedb then freezes.

Has anyone else seen this behaviour?

Martin Bowes

6. [Info-ingres] SIGSEGV in errlog.log

7. [Info-ingres] Iidbms crashing SIGBUS/SIGSEGV ; E_DM002B

8. [Info-Ingres] genxml and a WTF SIGSEGV



Return to DATABASE

 

Who is online

Users browsing this forum: No registered users and 97 guest