[jboss-jira] [JBoss JIRA] (JGRP-823) TCCPING: "failed creating initial list of hosts" fails stack startup

[Antonio Parolini (JIRA)]

    [ https://issues.jboss.org/browse/JGRP-823?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12805469#comment-12805469 ] 

Antonio Parolini commented on JGRP-823:
---------------------------------------

This is a big issue for us. 

If the DNS lookup fails anytime after that instance started, the TCPping protocol just stops working for ever, until you restart all the clustered nodes.

Here is what we got this morning:

{code}
2013-09-18 07:15:54,254 ERROR [org.jgroups.protocols.TCPGOSSIP] failed creating initial hosts
java.net.UnknownHostException: gossip-digprod-10: gossip-digprod-10
	at java.net.InetAddress.getAllByName0(InetAddress.java:1145)
	at java.net.InetAddress.getAllByName(InetAddress.java:1072)
	at java.net.InetAddress.getAllByName(InetAddress.java:1008)
	at java.net.InetAddress.getByName(InetAddress.java:958)
	at org.jgroups.stack.IpAddress.<init>(IpAddress.java:49)
	at org.jgroups.protocols.TCPGOSSIP.createInitialHosts(TCPGOSSIP.java:188)
	at org.jgroups.protocols.TCPGOSSIP.setProperties(TCPGOSSIP.java:71)
	at org.jgroups.stack.Protocol.setPropertiesInternal(Protocol.java:223)
	at org.jgroups.stack.Configurator$ProtocolConfiguration.createLayer(Configurator.java:707)
	at org.jgroups.stack.Configurator$ProtocolConfiguration.access$000(Configurator.java:606)
	at org.jgroups.stack.Configurator.createProtocols(Configurator.java:387)
	at org.jgroups.stack.Configurator.setupProtocolStack(Configurator.java:60)
	at org.jgroups.stack.ProtocolStack.setup(ProtocolStack.java:180)
	at org.jgroups.JChannel.open(JChannel.java:525)
	at org.jgroups.JChannel$CloserThread.run(JChannel.java:1635)
2013-09-18 07:15:54,255 ERROR [org.jgroups.JChannel] failure reopening channel: org.jgroups.ChannelException: failed to open channel
2013-09-18 07:17:52,867 ERROR [org.jgroups.stack.ProtocolStack] no down protocol available !
{code}
                
> TCCPING: "failed creating initial list of hosts" fails stack startup
> --------------------------------------------------------------------
>
>                 Key: JGRP-823
>                 URL: https://issues.jboss.org/browse/JGRP-823
>             Project: JGroups
>          Issue Type: Bug
>    Affects Versions: 2.6.3
>         Environment: RHEL4 x64, Sun JDK 1.5.0_16 64-bit
>            Reporter: P Eger
>            Assignee: Bela Ban
>             Fix For: 2.7
>
>         Attachments: stack.xml
>
>
> Using TCPPING is dangerous, as if any of the servers do not have a valid DNS entry the stack will fail with below execption. I would thing a much preferable behaviour in this case would be to silently (or with WARN) ignore those servers that do not name resolve. In the below example, csvr1and csvr2 exist and are online/started, csvr3 is attempting to start and csvr4 is offline/non-existent. csvr3 fails since it cannot DNS resolve csvr4, even though csvr1,csvr2 are avaiable and ready to process. Basically i think that a DNS lookup error should be treated equivalent to a connection failure, possibly by delaying the name resolution until later when the actual connect is attempted.
> <TCPPING timeout="15000"
> initial_hosts="${jgroups.tcpping.initial_hosts:csvr1[17801],csvr2[17801],csvr3[17801],csvr4[17801]}}"
> port_range="1"
> num_initial_members="2"/>
> ERROR | failed creating initial list of hosts
> java.net.UnknownHostException: csvr4: csvr4
>         at java.net.InetAddress.getAllByName0(InetAddress.java:1145)
>         at java.net.InetAddress.getAllByName(InetAddress.java:1072)
>         at java.net.InetAddress.getAllByName(InetAddress.java:1008)
>         at java.net.InetAddress.getByName(InetAddress.java:958)
>         at org.jgroups.stack.IpAddress.<init>(IpAddress.java:53)
>         at org.jgroups.protocols.TCPPING.createInitialHosts(TCPPING.java:147)
>         at org.jgroups.protocols.TCPPING.setProperties(TCPPING.java:77)
>         at org.jgroups.stack.Protocol.setPropertiesInternal(Protocol.java:97)
>         at org.jgroups.stack.Configurator$ProtocolConfiguration.createLayer(Configurator.java:828)
>         at org.jgroups.stack.Configurator$ProtocolConfiguration.access$000(Configurator.java:724)
>         at org.jgroups.stack.Configurator.createProtocols(Configurator.java:505)
>         at org.jgroups.stack.Configurator.setupProtocolStack(Configurator.java:61)
>         at org.jgroups.stack.ProtocolStack.setup(ProtocolStack.java:285)
>         at org.jgroups.JChannel.init(JChannel.java:1506)
>         at org.jgroups.JChannel.<init>(JChannel.java:246)
>         at org.jgroups.JChannel.<init>(JChannel.java:184)

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira