[jira] [Commented] (TS-2757) deleting stale API logs always crashes

[James Peach (JIRA)]

[ 
https://issues.apache.org/jira/browse/TS-2757?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13984891#comment-13984891
 ] 

James Peach commented on TS-2757:
-

I think that this crash is caused by the silly dance that {{LogObjectManager}} 
does when reloading the logging configuration. I can reproduce similar problems 
by running moderate traffic while reloading the logging configuration a lot.

The only way to fix all of these issues is to add full reference counting to 
{{LogObject}}. I have a patch that implements that and it has been running the 
test above safely for a few hours now.

> deleting stale API logs always crashes
> --
>
> Key: TS-2757
> URL: https://issues.apache.org/jira/browse/TS-2757
> Project: Traffic Server
>  Issue Type: Bug
>  Components: Logging
>Affects Versions: 5.0.0
>Reporter: James Peach
>Assignee: James Peach
>  Labels: A
>
> If you use {{TSTextLogObjectCreate}} in a plugin, when those log objects are 
> deleted, the object manager crashes. It looks like {{_APIobjects}} has a 
> stale pointer.
> {code}
> (gdb) bt
> #0  0x005e8c47 in LogObjectManager::~LogObjectManager 
> (this=0x4bf0028, __in_chrg=) at LogObject.cc:903
> #1  0x005d33e2 in LogConfig::~LogConfig (this=0x4bf, 
> __in_chrg=) at LogConfig.cc:573
> #2  0x005d3446 in LogConfig::~LogConfig (this=0x4bf, 
> __in_chrg=) at LogConfig.cc:573
> #3  0x00617b8a in ConfigProcessor::release (this=0xafa720, id=5, 
> info=0x4bf) at ProxyConfig.cc:210
> #4  0x00618496 in ConfigInfoReleaser::handle_event (this=0x21c5a00) 
> at ProxyConfig.cc:106
> #5  0x004e5920 in Continuation::handleEvent (this=0x21c5a00, event=2, 
> data=0xe926780) at ../iocore/eventsystem/I_Continuation.h:146
> #6  0x006d17b6 in EThread::process_event (this=0x2e08000, 
> e=0xe926780, calling_code=2) at UnixEThread.cc:145
> #7  0x006d1ad1 in EThread::execute (this=0x2e08000) at 
> UnixEThread.cc:224
> #8  0x006d0da0 in spawn_thread_internal (a=0x21ddd60) at Thread.cc:88
> #9  0x2b507e273851 in start_thread () from /lib64/libpthread.so.0
> {code}



--
This message was sent by Atlassian JIRA
(v6.2#6252)

[jira] [Commented] (TS-2757) deleting stale API logs always crashes

[Leif Hedstrom (JIRA)]

[ 
https://issues.apache.org/jira/browse/TS-2757?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13982672#comment-13982672
 ] 

Leif Hedstrom commented on TS-2757:
---

[~jpe...@apache.org] Is this a v5.0.0 fix ?

> deleting stale API logs always crashes
> --
>
> Key: TS-2757
> URL: https://issues.apache.org/jira/browse/TS-2757
> Project: Traffic Server
>  Issue Type: Bug
>  Components: Logging
>Reporter: James Peach
>  Labels: A
>
> If you use {{TSTextLogObjectCreate}} in a plugin, when those log objects ate 
> deleted, the object manager crashes. It looks like {{_APIobjects}} has a 
> stale pointer.
> {code}
> (gdb) bt
> #0  0x005e8c47 in LogObjectManager::~LogObjectManager 
> (this=0x4bf0028, __in_chrg=) at LogObject.cc:903
> #1  0x005d33e2 in LogConfig::~LogConfig (this=0x4bf, 
> __in_chrg=) at LogConfig.cc:573
> #2  0x005d3446 in LogConfig::~LogConfig (this=0x4bf, 
> __in_chrg=) at LogConfig.cc:573
> #3  0x00617b8a in ConfigProcessor::release (this=0xafa720, id=5, 
> info=0x4bf) at ProxyConfig.cc:210
> #4  0x00618496 in ConfigInfoReleaser::handle_event (this=0x21c5a00) 
> at ProxyConfig.cc:106
> #5  0x004e5920 in Continuation::handleEvent (this=0x21c5a00, event=2, 
> data=0xe926780) at ../iocore/eventsystem/I_Continuation.h:146
> #6  0x006d17b6 in EThread::process_event (this=0x2e08000, 
> e=0xe926780, calling_code=2) at UnixEThread.cc:145
> #7  0x006d1ad1 in EThread::execute (this=0x2e08000) at 
> UnixEThread.cc:224
> #8  0x006d0da0 in spawn_thread_internal (a=0x21ddd60) at Thread.cc:88
> #9  0x2b507e273851 in start_thread () from /lib64/libpthread.so.0
> {code}



--
This message was sent by Atlassian JIRA
(v6.2#6252)