accumulo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From dlmarion <dlmar...@comcast.net>
Subject RE: One instance, multiple Accumulo versions?
Date Mon, 09 Mar 2015 02:03:14 GMT
Typically I use different instances. I have some maven profiles[1] that I have been using for
functional testing, and I believe Keith has something that they use for Fluo.

[1] https://github.com/dlmarion/AccuFunk



<div>-------- Original message --------</div><div>From: Dylan Hutchison
<dhutchis@mit.edu> </div><div>Date:03/08/2015  9:47 PM  (GMT-05:00) </div><div>To:
Accumulo Dev List <dev@accumulo.apache.org> </div><div>Cc:  </div><div>Subject:
One instance, multiple Accumulo versions? </div><div>
</div>Are there guidelines on how to test different versions of Accumulo server
code for the same instance?  Or is the way to do it to use separate
instances, i.e., separate directories within Hadoop, for each version of
Accumulo we want to test?

I have been using the 1.6.1 release on my local computer and switched to
building from the trunk (1.7.0-SNAPSHOT) today, in order to see if an issue
is already fixed in the trunk.  I copied the conf/ files from my 1.6.2
build to the trunk build and couldn't get Accumulo to start properly.

The error is

2015-03-08 21:30:07,542 [server.Accumulo] FATAL: Problem verifying Fate
readiness
org.apache.accumulo.core.client.AccumuloException: Aborting upgrade because
there are *outstanding FATE transactions from a previous Accumulo version*.
Please see the README document for instructions on what to do under your
previous version.
        at
org.apache.accumulo.server.Accumulo.abortIfFateTransactions(Accumulo.java:317)

I then stopped the trunk Accumulo, started and cleanly stopped the 1.6.2
version (all fine), started the trunk version again and same problem.  So I
don't think the problem is stopping Accumulo with outstanding transactions
on the 1.6.2 version.  I looked for the documentation referenced in the
error on upgrading from a previous Accumulo version but could not find it.

Maybe the easiest thing to do is run the trunk version on a separate
instance.  Ideas?  Hadoop version is 2.6.0 and I build the native map, in
case that helps.

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