logging-log4net-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Reineri, Jim" <jrein...@prenova.com>
Subject RE: Coding horror deadlock
Date Fri, 05 Dec 2008 13:47:01 GMT
This is the situation described in Coding Horror:
        On thread #1, our code was doing Log (lock) / DB stuff (lock)
        On thread #2, our code was doing DB stuff (lock) / log stuff (lock)

The is a classic deadlock example.  If thread #1 locks 'Log' at the same time that thread
#2 locks 'DB Stuff' then thread #1 will be waiting for the 'DB stuff' lock to be freed before
it can complete its work and thread #2 will be waiting for the 'Log' lock to be freed before
completing its work.  Because each is waiting on the other, neither will ever release its
lock.

It is more likely to occur on heavily loaded, multi-core systems.

This problem is easily avoided by making sure that any code that needs both resources applies
the locks in the same order.

As was pointed out in the article, the problem was not a problem with log4net.  It really
is not a even problem with logging.  The problem was really bad code.

The author does make some good points about logging.  While I do not necessarily agree with
everything he said, it is worth a read.

Cheers,
Jim





-----Original Message-----
From: koger [mailto:karsten@f-kousgaard.dk]
Sent: Friday, December 05, 2008 6:56 AM
To: log4net-user@logging.apache.org
Subject: Coding horror deadlock


Hi

Those of you who reads coding horror, might have seen this post:
http://www.codinghorror.com/blog/archives/001192.html.

Could somebody explain me how this deadlock happend? And how do I avoid to
get any deadlocks myself using log4net?
--
View this message in context: http://www.nabble.com/Coding-horror-deadlock-tp20852380p20852380.html
Sent from the Log4net - Users mailing list archive at Nabble.com.


Mime
View raw message