#7 APPARENT DEADLOCK!!

[jfjimenezp]

I'm developing an application using Oracle 8.1.7, Sun
One Studio 5 and Hibernate 2.1.8 and c3p0-0.8.4.5.jar

While running the application I get the following error

com.mchange.v2.async.ThreadPoolAsynchronousRunne
r$DeadlockDetector@5acf13 -- APPARENT
DEADLOCK!!

The properties included in the hibernate.properties file
are :

hibernate.c3p0.min_size=5
hibernate.c3p0.max_size=20
hibernate.c3p0.timeout=300
hibernate.c3p0.max_statements=50
hibernate.c3p0.idle_test_period=3000
hibernate.c3p0.acquire_increment=2
hibernate.c3p0.validate=false

What is the cause?

Thanks.

[Daniel Edberg]

Logged In: YES
user_id=1269101

Hi,
I'm dealing with exactly the same problem. I use
c3p0-0.8.4.5, Hibernate3, mssql2000 and oracle 9.2i

the deadlock seems to occur under inactivity when the system
is idle for some time (couple of hours).. I havn't been able
to recreate the deadlock exactly when I want though

theese are my properties for c3p0:
hibernate.c3p0.max_size = 5
hibernate.c3p0.min_size = 2
hibernate.c3p0.timeout = 1800
hibernate.c3p0.max_statements = 100

this is a stacktrace of the deadlocking threads:

JProbe: Found Deadlock 1 (2 threads)
Thread
"com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#2"
(ID: 0x1a4acc10) blocked by thread
"com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#1"
(ID: 0x19147e28) at:
boolean
com.mchange.v1.db.sql.StatementUtils.attemptClose(java.sql.Statement)
void com.mchange.v2.c3p0.stmt.GooGooStatementCache$1.run()
void java.lang.Runnable.run()
void
com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread.run()
Thread
"com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#1"
(ID: 0x19147e28) blocked by thread
"com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#2"
(ID: 0x1a4acc10) at:
void java.sql.Connection.close()
void
com.mchange.v2.c3p0.impl.C3P0PooledConnection.close(boolean)
void com.mchange.v2.c3p0.impl.C3P0PooledConnection.close()
void javax.sql.PooledConnection.close()
void
com.mchange.v2.c3p0.impl.C3P0PooledConnectionPool$1.destroyResource(java.lang.Object)
void
com.mchange.v2.resourcepool.ResourcePool$Manager.destroyResource(java.lang.Object)
void com.mchange.v2.resourcepool.BasicResourcePool$5.run()
void java.lang.Runnable.run()
void
com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread.run()

in seriously need of a fix.. have you tried to find the
root of the deadlock problem?

Regards / Daniel Edberg

[Nobody/Anonymous]

Logged In: NO

I posted the same question to by Steve Waldman
<swaldman@mchange.com>, I think he belongs to the
developer's team, and he recommend me to upgrade to c3p0-
0.9.0-pre5 and that's what I did.

This versin has a better performace but although from time
to time appears the same trace, mainly when the application
is idle, the application doesn't hang anymore.

That's all for now.
I hope this helps you.

[Steve Waldman]

Logged In: YES
user_id=175530

Hi. Please give c3p0-0.9.0-pre6 a shot. I think it may resolve this issue. (If
I understand the problem correctly, some drivers, most notably Oracle,
deadlocked if Connections and their child Statements are concurrently and
simultaneously closed. c3p0 now ensures that cached
PreparedStatements closes are complete prior to calling Connection close.
Note that other drivers threw intermittent NullPointerExceptions under the
same or similar conditions, and these should also be fixed.)

I would appreciate feedback as to whether this fix actually works for y'all,
as this is not a problem I can replicate on my own. Thanks!

[Nobody/Anonymous]

Logged In: NO

Hi,
I encountered the same problem.
I'm using:
c3p0 0.0.9 pre 6
Hibernate 2.1.7 - JBoss 3.2.6 - JTDS 1.1 - SQL Server 7.0

Here is the trace if it can help:

15:56:15,092 WARN [ThreadPoolAsynchronousRunner]
com.mchange.v2.async.ThreadPoolAsynchronousRunner$DeadlockDetector@12b
23ce -- APPARENT DEADLOCK!!! Complete Status:
[num_managed_threads: 3, num_active: 3; activeTasks:
com.mchange.v2.resour
cepool.BasicResourcePool$5@1e69671
(com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#0),
com.mchange.v2.res
ourcepool.BasicResourcePool$5@19f0c0e
(com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#2),
com.mchange.v2.
resourcepool.BasicResourcePool$5@d0a462
(com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#1);
pendingTasks:
com.mchange.v2.resourcepool.BasicResourcePool$AcquireTask@1d837fe,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@4415
07, com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@1c09545,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@87fa6d, com
.mchange.v2.c3p0.stmt.GooGooStatementCache$2@14a1bd1,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@683d27,
com.mchang
e.v2.c3p0.stmt.GooGooStatementCache$2@1870b40,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@f3a6eb,
com.mchange.v2.c3
p0.stmt.GooGooStatementCache$2@53bcf3,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@19a0f55,
com.mchange.v2.c3p0.stmt
.GooGooStatementCache$2@164a18e,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@1aa735f]

Regards,
Guillaume Chevallier

[Nobody/Anonymous]

Logged In: NO

from gchevallier@rsa.fr

Hi,
I'm dealing with the same problem.
I'm using c3p0 0.0.9 pre 6 with:
Hibernate 2.1.7, JBoss 3.2.6, jtds 1.1, SQL Server 7.0

Here is the trace if it can help:
15:56:15,092 WARN [ThreadPoolAsynchronousRunner]
com.mchange.v2.async.ThreadPoolAsynchronousRunner$DeadlockDetector@12b
23ce -- APPARENT DEADLOCK!!! Complete Status:
[num_managed_threads: 3, num_active: 3; activeTasks:
com.mchange.v2.resour
cepool.BasicResourcePool$5@1e69671
(com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#0),
com.mchange.v2.res
ourcepool.BasicResourcePool$5@19f0c0e
(com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#2),
com.mchange.v2.
resourcepool.BasicResourcePool$5@d0a462
(com.mchange.v2.async.ThreadPoolAsynchronousRunner$PoolThread-#1);
pendingTasks:
com.mchange.v2.resourcepool.BasicResourcePool$AcquireTask@1d837fe,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@4415
07, com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@1c09545,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@87fa6d, com
.mchange.v2.c3p0.stmt.GooGooStatementCache$2@14a1bd1,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@683d27,
com.mchang
e.v2.c3p0.stmt.GooGooStatementCache$2@1870b40,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@f3a6eb,
com.mchange.v2.c3
p0.stmt.GooGooStatementCache$2@53bcf3,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@19a0f55,
com.mchange.v2.c3p0.stmt
.GooGooStatementCache$2@164a18e,
com.mchange.v2.c3p0.stmt.GooGooStatementCache$2@1aa735f]

Regards,
Guillaume Chevallier <gchevallier@rsa.fr>

[Venkat]

Logged In: YES
user_id=1305529

Hi Steve,

Howdy. I still get this same warning about deadlock with
MYSql DB and pool config below.

minPoolSize = 50
maxPoolSize =100
acquireIncrement = 5

Rest are defaults.

Any idea is greatly appreciated.

Thanks,
Venkatesh

[Jeetendra Mirchandani]

Logged In: YES
user_id=877396

I use c3p0 0.9.0.2 with mysql and am facing the same problem.
Any clues?

[Steve Waldman]

Logged In: YES
user_id=175530

Hi.

Can you get a thread dump from a VM after (preferably soon after) you see this
message? Also, please include the active and pending tasks lists that get
dumped with this message. Thanks!

And sorry to all for a long absence!

[Steve Waldman]

Logged In: YES
user_id=175530

Hi. I think, at long last, this problem is finally fixed.
Daniel Edberg's jprobe stack trace points directly to the
problem. Unfortunately, somehow I missed that, and let a
year go by before finding this same deadlock myself. Thank
you Daniel for the excellent sleuthing, and sorry it has
been the fix has been so long in coming. If any users who've
experience the APPARENT DEADLOCK issues can try out
c3p0-0.9.1-pre7 and let me know whether they seem to be
resolved, I'd appreciate that.