valgrind [logs and help]

[Wyk3d]

sd2 857(modified some c++ scripts sorry)

There are a LOT of serious SD2 errors! They could be causing some of the core bugs/crashes/freezes but it's impossible to fix them if you don't use clean SD2.

i also reproduced crash described here:

Did you reproduce it while running under valgrind ? The first log ends with a freeze and the second one seems to be shut down/terminated.

[painejake]

Latest crash log on Win Server 32 in Debug: http://paste2.org/p/135233

Tried purify and it said i need to have the full version to run it. So i will try a different program later

[painejake]

Ok under Debian 1.6 Days uptime and counting peaks of 150 with no lag at all. MaNGOS rev 6928 SD rev 812

Edit: Crashed after 2.1 days and peak of 180 testers. Around 120 on 24/7

[Wyk3d]

sorry for BIG delay, i didnt had mutch time to do some logs, so here are promised logs, i can provide more if need ... hope it help, i would really like to see some fixes from thees logs on mangos GIT so i know it helped

There were already commits fixing problems from your previous logs. I think about 2 crashes and memory leaks, don't remember exactly. They also helped confirm another major problem with grid unloading though the fix for that is yet to be written.

[oc_redfox]

There were already commits fixing problems from your previous logs. I think about 2 crashes and memory leaks, don't remember exactly. They also helped confirm another major problem with grid unloading though the fix for that is yet to be written.

Major problem with Grid unloading? seems familiar ^^...

[higi]

did many more test, with :

tested on : debug mode, no optimatization+(some flags for quade core)

Mangos/0.13.0 (**REvision 7723-* ) for Unix ( little-endian)

Using script library: ScriptDev2 (for MaNGOS 7715+) Revision [1029] 2009-04-26 23:45:06 (Unix)

Using World DB: UDB 0.11.0 (371) for MaNGOS 7681 with SD2 SQL for wev. 1012

Using creature EventAI: ACID 0.0.9 - Full Release for Mangos 7673+

CLEAN MANGOS + CLEAN SD2 + LOS/HEIGH ON(ALL MAPS)

here are many valgrind logs, we got some days testing, so hope it will help to find new crash/leaks..

if devs will need bt full for some logs, just send me log numbers, each log mean 1 crash

http://filebeam.com/91e5b69e07765701972aedcf131b136e

or any extra flags for valgrind setting, or just say what DEVs will need.

if anything helped, tell me pls:)

[dereka]

did many more test, with :

tested on : debug mode, no optimatization+(some flags for quade core)

Mangos/0.13.0 (**REvision 7723-* ) for Unix ( little-endian)

Using script library: ScriptDev2 (for MaNGOS 7715+) Revision [1029] 2009-04-26 23:45:06 (Unix)

Using World DB: UDB 0.11.0 (371) for MaNGOS 7681 with SD2 SQL for wev. 1012

Using creature EventAI: ACID 0.0.9 - Full Release for Mangos 7673+

CLEAN MANGOS + CLEAN SD2 + LOS/HEIGH ON(ALL MAPS)

here are many valgrind logs, we got some days testing, so hope it will help to find new crash/leaks..

if devs will need bt full for some logs, just send me log numbers, each log mean 1 crash

http://filebeam.com/91e5b69e07765701972aedcf131b136e

or any extra flags for valgrind setting, or just say what DEVs will need.

if anything helped, tell me pls:)

thank you!

[aaded]

big thanks:)

fixed one memory leak

http://bit.ly/4fAtT

3 crash fixed by Vladimir in [7769] [7770] and [7771] rev

2 crash may be fixed by NoFantasy in SD2

maybe you can try to use valgrind to the new rev?

and without stiil reachable errors

--show-reachable=no

[higi]

awesome, didnt think it can help that mutch, thank u for infos, now i really know its need...)

i just did clean build 7782 + latest sd2

will post new logs in morning, after we collect new data, and i did that options u suggest.

EDIT: have problem with our test server, cant connect to it, so logs will be later, i need to get server startup

[NoFantasy]

Just popping in to say thanks a bunch for this Higi! I admit it's a pain to go through the logs, try understand, and of course attempt to fix where should be fixed but i clearly see it's worth every bits doing this.

[Auntie Mangos]

should this be:

Total Disc space = size of log pages number of log pages per log max number of primary logs max number of secondary logs..?

Yes, probably.

[lillecarl]

[VladimirMangos]

Hmm, as i look most reports about destructor calling for global singletons. valgrind can incorrect false detect problems maybe...

Because related fields cleanup in code in singleton destructors...

BattleGroundMgr::DeleteClientVisibleInstanceId case uninitilized conditinal case strange but i will change cdeo: it not need check in this point.