-Re: tserver error for accumulo v1.5
John Vines 2013-08-30, 17:42
The message indicates that the lock was deleted, which means there should
be a corresponding message on the master for deleting that lock. More
specifically, the master attempts to contact the tserver and if it fails to
do so, it will delete the lock so the tserver dies. When the tserver comes
up, it will report it's IP address in zookeeper to acquire the lock, so if
it's for some reason reporting a bad IP address, this could be the reason.
I would start with checking the corresponding master message and making
sure that the IP address for the lock it's deleting is valid.
On Fri, Aug 30, 2013 at 1:35 PM, Chen, Yinuo J CIV SPAWARSYSCEN-PACIFIC,
55390 <[EMAIL PROTECTED]> wrote:
> I have an accumulo error regarding tservers not coming up correctly. The
> tserver only comes up on one machine and it is the same machine as the
> accumulo master.
> The error that the tserver log file gave me is:
> FATAL: Lost tablet server lock (reason LOCK_DELETED), exiting.
> From this, I deduced it might be a communication error between the tserver
> and zookeeper. However, current operations is not load intensive, so I
> don't know why it wouldn't communicate. There is nothing in the .err and
> .out file.
> My current setup is as followed:
> Apache Accumulo 1.5
> Apache Hadoop 1.1.2
> Apache Zookeeper 3.4.5
> This never happened with Accumulo 1.4.x, so maybe this is a new
> configuration that I might have glossed over.=20
> If there is any information regarding why this tserver won't come up, I
> would appreciate it.
> James Chen