nifi-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Phil H <gippyp...@gmail.com>
Subject RE: Zookeeper - help!
Date Wed, 03 Oct 2018 07:09:18 GMT
Okay,

I have got this working now, albeit with only a single ZK instance (at this stage).

The missing piece of the puzzle that wasn’t in the guides from Pierre was that cluster servers’
certificates need to be installed in each server’s keystore, and all the cluster server
DNs need to be added as Initial User Identities in authorizers.xml.

Thanks again for all the assistance.

Sent from Mail for Windows 10

From: Nathan Gough
Sent: Wednesday, 3 October 2018 7:27 AM
To: dev@nifi.apache.org
Subject: Re: Zookeeper - help!

I think you are correct on that, I assumed it was a range of some kind but it looks like it's
not: http://zookeeper.apache.org/doc/r3.4.3/zookeeperStarted.html#sc_RunningReplicatedZooKeeper


On 10/2/18, 5:17 PM, "Phil H" <gippyphil@gmail.com> wrote:

    The second port in the zookeeper server config has been a mystery to me.  I thought it
was a second port used for elections, not the upper bound in a range.  Why is the range so
large?
    
    Sent from Mail for Windows 10
    
    From: Nathan Gough
    Sent: Wednesday, 3 October 2018 1:26 AM
    To: dev@nifi.apache.org
    Subject: Re: Zookeeper - help!
    
    Check your configs on nifi2. I don't believe that NiFi is starting two instances of Zookeeper
but the ports configured are unintentionally configured to overlap ie. Ports used twice in
different configs where they should be different.
    
    It may be that your zookeeper.properties has:
    
    clientPort=2180
    ...
    server.1=nifi1.com:2180:3888
    server.2=nifi2.com:2180:3888
    
    where it should be:
    
    clientPort=2180
    ...
    server.1=nifi1.com:2888:3888
    server.2=nifi2.com:2888:3888
    
    noticing that the server.1 and server.2 ranges don't overlap with the client port.
    
    
    Not sure if this helps, but the following is the relevant config that I have for my NiFi
cluster nodes that run on the SAME machine where nifi1.com and nifi2.com are configured in
/etc/hosts:
    
    nifi1/conf
    zookeeper.properties
    - clientPort=2180
    - server.1=nifi1.com:2888:3888
    - server.2=nifi2.com:2888:3888
    
    nifi.properties
    - nifi.remote.input.host=nifi1.com
    - nifi.remote.input.socket.port=10440
    - nifi.web.http.host=nifi1.com
    - nifi.web.http.port=9550
    - nifi.cluster.node.address=nifi1.com
    - nifi.cluster.node.protocol.port=11440
    
    nifi1/state/zookeeper
    /myid (file contents = "1")
    /state-management.xml (no changes required)
    /version-2/
    
    
    nifi2/conf
    zookeeper.properties
    - clientPort=2181
    - server.1=nifi1.com:2888:3888
    - server.2=nifi2.com:2888:3888
    
    nifi.properties
    - nifi.remote.input.host=nifi2.com
    - nifi.remote.input.socket.port=10441
    - nifi.web.http.host=nifi2.com
    - nifi.web.http.port=9551
    - nifi.cluster.node.address=nifi2.com
    - nifi.cluster.node.protocol.port=11441
    
    nifi2/state/zookeeper
    /myid (file contents = "2")
    /state-management.xml (no changes required)
    /version-2/
    
    
    Nathan
    
    
    
    On 10/2/18, 2:07 AM, "Phil H" <gippyphil@gmail.com> wrote:
    
        Hi Andy,
        
        Thanks for the additional info.  I think I saw a link to that while searching but
was wary since it was such an old version.
        
        I have two VMs (nifi1, and nifi2) both running NiFi with identical configs, and trying
to use the inbuilt ZK to cluster them.
        
        If I only mention a single machine within the config (eg: if nifi1 doesn’t refer
to nifi2, or visa versa) I don’t get any start up errors.
        
        Phil
        
        From: Andy LoPresto
        Sent: Tuesday, 2 October 2018 1:00 PM
        To: dev@nifi.apache.org
        Subject: Re: Zookeeper - help!
        
        Hi Phil, 
        
        Nathan’s advice is correct but I think he was assuming all other configurations
are correct as well. Are you trying to run both NiFi nodes and ZK instances on the same machine?
In that case you will have to ensure that the ports in use are different for each service
so they don’t conflict. Setting them all to the same value only works if each service is
running on an independent physical machine, virtual machine, or container. 
        
        I find Pierre’s guide [1] to be a helpful step-by-step instruction list as well
as a good explanation of how the clustering concepts work in practice. When you get that working,
and you’re ready to set up a secure cluster, he has a follow-on guide for that as well [2].
Even as someone who has set up many clustered instances of NiFi, I use his guides regularly
to ensure I haven’t forgotten a step. 
        
        They were originally written for versions 1.0.0 and 1.1.0, but the only thing that
has changed is the authorizer configuration for the secure instances (you’ll need to put
the Initial Admin Identity and Node Identities in two locations in the authorizers.xml file
instead of just once). 
        
        Hopefully this helps you get a working cluster up and running so you can experiment.
Good luck. 
        
        [1] https://pierrevillard.com/2016/08/13/apache-nifi-1-0-0-cluster-setup/
        [2] https://pierrevillard.com/2016/11/29/apache-nifi-1-1-0-secured-cluster-setup/
        
        
        Andy LoPresto
        alopresto@apache.org
        alopresto.apache@gmail.com
        PGP Fingerprint: 70EC B3E5 98A6 5A3F D3C4  BACE 3C6E F65B 2F7D EF69
        
        On Oct 1, 2018, at 2:45 PM, Phil H <gippyphil@gmail.com> wrote:
        
        Thanks Nathan,
        
        I changed the protocol.port to 10002 on both servers.
        
        On server 1, I now just see endless copies of the second error from my original message
(“KeeperException$ConnectionLossException: KeeperErrorCode = ConnectionLoss”) – I don’t
know if that’s normal when there’s only a single member of a cluster alive and running?
 Seems like the logs will fill up very quickly if it is!
        
        On server 2, I get a bind exception on the Zookeeper client port.  It doesn’t matter
what I set it to (In this example, I changed it to 10500) I always get the same result.  If
I run netstat when nifi isn’t running, there’s nothing listening on the port.  It’s
like NiFi is starting two Zookeeper instances?!  There’s no repeat of this in the start
up sequence though.  Both servers are running completely vanilla 1.6.0 – I don’t even
have any flow defined yet, this is purely for teaching myself clustering config – so I don’t
know why one is behaving differently to the other.
        
        2018-10-02 17:36:31,610 INFO [QuorumPeer[myid=2]/0.0.0.0:10500] o.a.zookeeper.server.ZooKeeperServer
Created server with tickTime 2000 minSessionTimeout 4000 maxSessionTimeout 40000 datadir ./state/zookeeper/version-2
snapdir ./state/zookeeper/version-2
        2018-10-02 17:36:31,612 ERROR [QuorumPeer[myid=2]/0.0.0.0:10500] o.apache.zookeeper.server.quorum.Leader
Couldn't bind to nifi2.domain/192.168.10.102:10500
        java.net.BindException: Address already in use (Bind failed)
        	at java.net.PlainSocketImpl.socketBind(Native Method)
        	at java.net.AbstractPlainSocketImpl.bind(AbstractPlainSocketImpl.java:387)
        	at java.net.ServerSocket.bind(ServerSocket.java:375)
        	at java.net.ServerSocket.bind(ServerSocket.java:329)
        	at org.apache.zookeeper.server.quorum.Leader.<init>(Leader.java:193)
        	at org.apache.zookeeper.server.quorum.QuorumPeer.makeLeader(QuorumPeer.java:605)
        	at org.apache.zookeeper.server.quorum.QuorumPeer.run(QuorumPeer.java:798)
        
        
        
        
        From: Nathan Gough
        Sent: Tuesday, 2 October 2018 2:22 AM
        To: dev@nifi.apache.org
        Subject: Re: Zookeeper - help!
        
        Hi Phil,
        
        One thing I notice with your config is that the cluster.node.protol.port and the zookeeper
ports are the same - these should not be the same. Node.protocol.port is used by NiFi cluster
to communicate between nodes, the zookeeper.connect.string port should be the port that zookeeper
service is listening on. The zookeeper port is configured by the clientPort property in the
zookeeper.properties file. This would make your connect string: 'nifi.zookeeper.connect.string=nifi1.domain:2180,nifi2.domain:2180',
where 2180 is whatever clientPort is configured.
        
        You can read more about how NiFi uses Zookeeper and how to configure it here: https://nifi.apache.org/docs/nifi-docs/html/administration-guide.html#state_management.
        
        Let us know what happens once these properties are configured correctly.
        
        Nathan
        
        
        On 9/30/18, 11:07 PM, "Phil H" <gippyphil@gmail.com> wrote:
        
           Hi guys,
        
           Pulling my hair out trying to solve my Zookeeper problems.  I have two 1.6.0 servers
that I am trying to cluster.
        
           Here is the except from the properties files – all other properties are default
so omitted for clarity.   The servers are set up to run HTTPS, and the interface works via
the browser, so I believe the certificates are correctly installed.
        
           Server nifi1.domain:
           nifi.cluster.is.node=true
           nifi.cluster.node.address=nifi1.domain
           nifi.cluster.node.protocol.port=10000
        
           nifi.zookeeper.connect.string=nifi2.domain:10000,nifi1.domain:10000
           nifi.zookeeper.root.node=/nifi
        
           Server nifi2.domain:
           nifi.cluster.is.node=true
           nifi.cluster.node.address=nifi2.domain
           nifi.cluster.node.protocol.port=10000
        
           nifi.zookeeper.connect.string=nifi1.domain:10000,nifi2.domain:10000
           nifi.zookeeper.root.node=/nifi
        
           I am getting these errors (this is from server 2, but seeing the same on server
1 apart from a different address, of course):
        
           2018-10-01 20:54:16,332 INFO [main] org.apache.nifi.io.socket.SocketListener Now
listening for connections from nodes on port 10000
           2018-10-01 20:54:16,381 INFO [main] o.apache.nifi.controller.FlowController Successfully
synchronized controller with proposed flow
           2018-10-01 20:54:16,435 INFO [main] o.a.nifi.controller.StandardFlowService Connecting
Node: nifi2.domain:443
           2018-10-01 20:54:16,769 ERROR [Process Cluster Protocol Request-1] o.a.nifi.security.util.CertificateUtils
The incoming request did not contain client certificates and thus the DN cannot be extracted.
Check that the other endpoint is providing a complete client certificate chain
           2018-10-01 20:54:16,771 WARN [Process Cluster Protocol Request-1] o.a.n.c.p.impl.SocketProtocolListener
Failed processing protocol message from nifi2 due to org.apache.nifi.cluster.protocol.ProtocolException:
java.security.cert.CertificateException: javax.net.ssl.SSLPeerUnverifiedException: peer not
authenticated
           org.apache.nifi.cluster.protocol.ProtocolException: java.security.cert.CertificateException:
javax.net.ssl.SSLPeerUnverifiedException: peer not authenticated
                   at org.apache.nifi.cluster.protocol.impl.SocketProtocolListener.getRequestorDN(SocketProtocolListener.java:225)
                   at org.apache.nifi.cluster.protocol.impl.SocketProtocolListener.dispatchRequest(SocketProtocolListener.java:131)
                   at org.apache.nifi.io.socket.SocketListener$2$1.run(SocketListener.java:136)
                   at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
                   at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
                   at java.lang.Thread.run(Thread.java:748)
           Caused by: java.security.cert.CertificateException: javax.net.ssl.SSLPeerUnverifiedException:
peer not authenticated
                   at org.apache.nifi.security.util.CertificateUtils.extractPeerDNFromClientSSLSocket(CertificateUtils.java:314)
                   at org.apache.nifi.security.util.CertificateUtils.extractPeerDNFromSSLSocket(CertificateUtils.java:269)
                   at org.apache.nifi.cluster.protocol.impl.SocketProtocolListener.getRequestorDN(SocketProtocolListener.java:223)
                   ... 5 common frames omitted
           Caused by: javax.net.ssl.SSLPeerUnverifiedException: peer not authenticated
                   at sun.security.ssl.SSLSessionImpl.getPeerCertificates(SSLSessionImpl.java:440)
                   at org.apache.nifi.security.util.CertificateUtils.extractPeerDNFromClientSSLSocket(CertificateUtils.java:299)
                   ... 7 common frames omitted
        
        
        
           2018-10-01 20:54:32,249 INFO [Curator-Framework-0] o.a.c.f.state.ConnectionStateManager
State change: SUSPENDED
           2018-10-01 20:54:32,250 ERROR [Curator-Framework-0] o.a.c.f.imps.CuratorFrameworkImpl
Background operation retry gave up
           org.apache.zookeeper.KeeperException$ConnectionLossException: KeeperErrorCode =
ConnectionLoss
                   at org.apache.zookeeper.KeeperException.create(KeeperException.java:99)
                   at org.apache.curator.framework.imps.CuratorFrameworkImpl.checkBackgroundRetry(CuratorFrameworkImpl.java:728)
                   at org.apache.curator.framework.imps.CuratorFrameworkImpl.performBackgroundOperation(CuratorFrameworkImpl.java:857)
                   at org.apache.curator.framework.imps.CuratorFrameworkImpl.backgroundOperationsLoop(CuratorFrameworkImpl.java:809)
                   at org.apache.curator.framework.imps.CuratorFrameworkImpl.access$300(CuratorFrameworkImpl.java:64)
                   at org.apache.curator.framework.imps.CuratorFrameworkImpl$4.call(CuratorFrameworkImpl.java:267)
                   at java.util.concurrent.FutureTask.run(FutureTask.java:266)
                   at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180)
                   at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
                   at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
                   at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
                   at java.lang.Thread.run(Thread.java:748)
        
        
        
        
        
        
        
    
    
    
    




Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message