OSDir


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Kahadb with leased database locker broker not closing connector when connection to NFS fails.


I'm not an expert on NFS, but I'd have expected that the network
connectivity failure to the NFS share to be detected when the broker
attempts to read or write the file, not just when you disable connections
to it. Your description implies that you're doing the latter; what is the
behavior of the broker when you do the former, e.g. by attempting to
publish a message to the broker after you disable the network connectivity
to the share?

Tim

On Mon, May 7, 2018 at 5:42 PM, Suvp <sunil.v_p@xxxxxxxxxxxxxxxxx> wrote:

> Hi,
>
> In our activemq installation we are using kahaDb as the persistence adapter
> with leased-database-locker for locking mechanism. The shared file system
> used is NFSv3.
>
> We wanted to test scenario where the Master node has access to mysql
> through
> which it will remain master by acquiring lock but looses connection to
> shared file system or NFS and see that the master switches since its not
> able to contact NFS.
>
> What i observed was in the activemq logs there are no exceptions when
> active
> broker looses connection to NFS and the it still remains as master and the
> clients fail to receive messages when it looses connection.
>
> Following is the Exception handler used:
>         <ioExceptionHandler>
>             <leaseLockerIOExceptionHandler ignoreSQLExceptions="false"
> stopStartConnectors="true"/>
>         </ioExceptionHandler>
>
> And following is the persistence adapter configuration:
>           <persistenceAdapter>
>     <kahaDB directory="/home/sunil" lockKeepAlivePeriod="5000">
>         <locker>
>
>             <lease-database-locker lockAcquireSleepInterval="10000"
> dataSource="#mysql-ds">
>                 <statements>
>
>                     <statements lockTableName="ACTIVEMQ_LOCK"/>
>                 </statements>
>             </lease-database-locker>
>         </locker>
>     </kahaDB>
> </persistenceAdapter>
>
>
> And the NFS configuration:
> 10.31.103.147:/home/sunil   20G  103M   20G   1% /home/sunil
>
> /home/sunil 10.31.52.55(rw,sync,fsid=0) 10.31.52.56(rw,sync,fsid=0)
> 10.31.52.57(rw,sync,fsid=0)
>
> From the NFS server using iptable rules the master node is dropped:
> 10.31.52.56 is master node
> iptables -I OUTPUT -s 10.31.52.56 -j DROP
> iptables -I INPUT -s 10.31.52.56 -j DROP
> service iptables save
>
> Does the activemq raise IOException when it looses connection to Shared
> Filesystem ? And i can use NFSv3 as the shared file system rite ?
>
> Thanks in advance
>
>
>
>
> --
> Sent from: http://activemq.2283324.n4.nabble.com/ActiveMQ-User-
> f2341805.html
>