Hi Hugh,

The crash happens after a few thousand POST requests to /sparql/ -- the chances of a simple test case are slim...

If you can provide some tips and debugging code I'm willing to investigate further on my system. I've looked at Dk/Dkalloc.c and added "assert(sz != 0) ;" into av_adjust() but this doesn't catch anything, so it's rather puzzling.


Thanks,
Dave


On 1/03/13 2:24 AM, Hugh Williams wrote:
Hi David,

Are you able to provide a simple test case for recreation as it seems to be consistently and easily recreate'able on your machine and we can try on a 10.6 or 10.8 Mac build here. I presume u have not tried on another OS like Linux or other ?

Best Regards
Hugh Williams
Professional Services
OpenLink Software, Inc.      // http://www.openlinksw.com/
Weblog   -- http://www.openlinksw.com/blogs/
LinkedIn -- http://www.linkedin.com/company/openlink-software/
Twitter  -- http://twitter.com/OpenLink
Google+  -- http://plus.google.com/100570109519069333827/
Facebook -- http://www.facebook.com/OpenLinkSoftware
Universal Data Access, Integration, and Management Technology Providers

On 28 Feb 2013, at 08:33, David Brooks <[email protected] <mailto:[email protected]>> wrote:

A further update -- the only debugging option needed to prevent the crash is --with-malloc-debug.


Thanks.
Dave

On 28/02/13 8:19 PM, David Brooks wrote:
An update:

 1. The divide-by-zero happens with the latest sources in git (0e12bf7)
 2. The divide-by-zero will repeatedly occur after a given sequence
    of queries against the store (having now loaded all datasets
    there is little network traffic).
 3. The divide-by-zero doesn't happen if the latest sources are
    built with debug options enabled (debug, malloc-debug and
    mtx-debug).


On 28/02/13 1:19 PM, David Brooks wrote:
On 28/02/13 1:04 PM, David Brooks wrote:
Hi,

I'm having VOS crash with "Server received signal 8. Continuing with the default action for that signal." when using the SPARQL endpoint (signal 8 is SIGFPE).

This is happening under network load -- another process is receiving and saving streamed datasets, storing associated metadata in Virtuoso via SPARQL. Apart from network activity, nothing is particularly loaded -- each dataset generates around 30 triples and there is a new dataset every 5 to 10 seconds.

The crash isn't dataset specific but is repeatable -- restarting Virtuoso, then cleaning up the last dataset (from when the system crashed) and continuing the stream gets things going again. After another couple of minutes or so Virtuoso crashes again...

VOS 06.01.3127 on OS/X Server 10.6

Following are details from the crash report:

Exception Type:  EXC_ARITHMETIC (SIGFPE)
Exception Codes: EXC_I386_DIV (divide by zero)
Crashed Thread:  24

Thread 24 Crashed:
0   virtuoso-t 0x00000001003c2a8e av_adjust + 110
1   virtuoso-t 0x00000001003c3526 dk_alloc + 454
2   virtuoso-t 0x0000000100189a84 sqlr_run_bif_in_sandbox + 84
3 virtuoso-t 0x00000001001a5b7d sqlp_patch_call_if_special_or_optimizable + 957
4   virtuoso-t 0x00000001000dc69e yyparse + 9118
5   virtuoso-t 0x0000000100151d4f sql_compile_1 + 4783
6   virtuoso-t 0x00000001000f6dfa bif_exec_score + 298
7   virtuoso-t 0x0000000100187113 ins_call_bif + 51
8   virtuoso-t 0x000000010018bc5f code_vec_run_1 + 1935
9   virtuoso-t 0x00000001001af14a qn_input + 106
10  virtuoso-t 0x00000001001b310b qr_subq_exec + 779
11  virtuoso-t 0x000000010018a2cf ins_call + 559
12  virtuoso-t 0x000000010018bbc9 code_vec_run_1 + 1785
13  virtuoso-t 0x00000001001af14a qn_input + 106
14  virtuoso-t 0x00000001001b310b qr_subq_exec + 779
15  virtuoso-t 0x000000010018a2cf ins_call + 559
16  virtuoso-t 0x000000010018bbc9 code_vec_run_1 + 1785
17  virtuoso-t 0x00000001001af14a qn_input + 106
18  virtuoso-t 0x00000001001b310b qr_subq_exec + 779
19  virtuoso-t 0x000000010018a2cf ins_call + 559
20  virtuoso-t 0x000000010018bbc9 code_vec_run_1 + 1785
21  virtuoso-t 0x00000001001af14a qn_input + 106
22  virtuoso-t 0x00000001001b421d qr_exec + 861
23  virtuoso-t 0x00000001001b4c58 qr_quick_exec + 456
24  virtuoso-t 0x000000010029c654 ws_dav + 772
25  virtuoso-t 0x000000010007691f ws_request + 3951
26  virtuoso-t 0x000000010007a331 ws_read_req + 1105
27  virtuoso-t 0x000000010007a4cf ws_serve_connection + 239
28  virtuoso-t 0x000000010007a899 ws_init_func + 105
29  virtuoso-t 0x00000001003da73f thread_exit + 591
30  libSystem.B.dylib 0x00007fff86888fd6 _pthread_start + 331
31  libSystem.B.dylib 0x00007fff86888e89 thread_start + 13




------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_feb


_______________________________________________
Virtuoso-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/virtuoso-users

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_feb_______________________________________________
Virtuoso-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/virtuoso-users


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_feb
_______________________________________________
Virtuoso-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/virtuoso-users

Reply via email to