A large memory leak

[jben94]

Hello,

We have a large memory leak issue regarding the "malloc" line 90 of file "packet.cc".
We lost 4MB, 8MB of RAM in 15 minutes of use.

I did not find in your memory the release of this code "malloc".
Will you help us fix this problem?

We found this problem by using the tool Valgrind.

[SpamapS]

Hi jben94! Sounds like it's worthy of some inspection. Can you share with us the exact valgrind command you used, what version of valgrind, and what platform you ran it on? THANKS!

[jben94]

Hi,
Here are the different versions:

• Valgrind 3.11.0, command used : valgrind -v --tool=memcheck --leak-check=full --show-leak-kinds=all --log-file=/var/log/valgrind.log
• Gearmand 1.1.11
• Mod-gearman 1.5.5
• Redhat 6.7 (Santiago)

Changes made to bypass the problem are in the file.

diff.gearmand.txt

[p-alik]

I did a dry run without patching

• ubuntu 16.04
• valgrind-3.11.0
• gearmand 1.1.13 (+ bccf801)

Except still reachable it looks fine to me. Here entire valgrind.log. Is there anything I overseen?

==25057== 
==25057== HEAP SUMMARY:
==25057==     in use at exit: 72,704 bytes in 1 blocks
==25057==   total heap usage: 2,066 allocs, 2,065 frees, 278,679 bytes allocated
==25057== 
==25057== Searching for pointers to 1 not-freed blocks
==25057== Checked 1,217,456 bytes
==25057== 
==25057== 72,704 bytes in 1 blocks are still reachable in loss record 1 of 1
==25057==    at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==25057==    by 0x6DA3EFF: ??? (in /usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.21)
==25057==    by 0x40104E9: call_init.part.0 (dl-init.c:72)
==25057==    by 0x40105FA: call_init (dl-init.c:30)
==25057==    by 0x40105FA: _dl_init (dl-init.c:120)
==25057==    by 0x4000CF9: ??? (in /lib/x86_64-linux-gnu/ld-2.23.so)
==25057==    by 0x4: ???
==25057==    by 0xFFF00000E: ???
==25057==    by 0xFFF00002A: ???
==25057==    by 0xFFF000031: ???
==25057==    by 0xFFF000036: ???
==25057==    by 0xFFF000040: ???
==25057== 
==25057== LEAK SUMMARY:
==25057==    definitely lost: 0 bytes in 0 blocks
==25057==    indirectly lost: 0 bytes in 0 blocks
==25057==      possibly lost: 0 bytes in 0 blocks
==25057==    still reachable: 72,704 bytes in 1 blocks
==25057==         suppressed: 0 bytes in 0 blocks
==25057== 
==25057== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)
==25057== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

[jben94]

how do you use gearman server?
this rise was actually how long to use?
From my side, I use the mod-gearman module centreon-engine.

[p-alik]

gearmand started in valgrind as described in your comment at Aug 22 and run simple sum test.
I didn't use mod-gearman. Maybe is mod-gearman to blame for the issue?

[jben94]

I do not have the results of my tests with valgrind. But if you look at the portion of source code that I indicated. We can see a memory allocation that is never freed.

[p-alik]

@jben94, mod-gearman supports gearmand v0.33.
See

• Supported Dependencies
• gearmand issues "Too many open files" in combination with mod-gearman sni/mod_gearman#105

Maybe it's a clue for the issue.