RE: I'll buy you dinner if you can help me with strace/java ETIMEDOUT challenge :)
Date: Fri, 12 Apr 2013 15:50:20 -0500
Message-ID: <F05D8DF1FB25F44085DB74CB916678E8879BE525FA_at_NADCWPMSGCMS10.hca.corpad.net>
The problem is more of a challenge which is: How to reduce overhead in the java engine for a heavily utilized java app that executes around 200 jdbc connections during a 4 hour batch window.
So the effort is to skinny down what java is doing by "not doing work that is not needed".
iptables is off for the database servers.
Chris
-----Original Message-----
From: Iliya Peregoudov [mailto:iperegudov_at_cboss.ru]
Sent: Friday, April 12, 2013 1:13 AM
To: Taylor Christopher - Nashville
Cc: oracle-l_at_freelists.org
Subject: Re: I'll buy you dinner if you can help me with strace/java ETIMEDOUT challenge :)
ETIMEDOUT errors in the strace output you've attached are all returned by futex() system call. This system call is not related to networking. futex() system call is used to implement POSIX threads locking primitives like mutex, condvar, etc. And ETIMEDOUT error is not actually an error for pthread_cond_timedwait() or similar.
It will be better if you tell the problem you have. Do you have connection timeout when connecting from App server to DB server? Network problems are better to diagnose using nslookup and traceroute. Personally I bet on a misconfigured iptables firewall on DB server.
On 11.04.2013 22:29, Christopher.Taylor2_at_parallon.net wrote:
> I'm getting a weird ETIMEDOUT error on a linux appserver connecting to my database server in the STRACE output and was curious if anyone has worked through anything like this?
> This is a JDBC connection using SUN (Oracle) java 1.7
>
> Googling this seems to be [maybe] related to:
> a.) IPV6 enabled class where IPV6 is not enabled on the networks
> involved
> b.) network subnets maybe diff between appserver and db server
-- http://www.freelists.org/webpage/oracle-lReceived on Fri Apr 12 2013 - 22:50:20 CEST