fluo-muchos
fluo-muchos copied to clipboard
apache
Apache Fluo Muchos
I tried running 2 tablet servers per node and had to add the following to accumulo-site.xml. Would be nice to just add these. ```xml tserver.port.search true replication.receipt.service.port 10002-10009 ```
The property `twill_reserve_mem_mb` in `muchos.props` no longer works to configure reserved memory. Currently a Twill property is set in `yarn-site.xml` which needs to be removed. Could configure fluo-yarn to use...
Currently the cluster stats plots CPU and load. It would be nice to plot disk I/O and network I/O rates also.
I setup a 10 node cluster and forgot to annotate a node for Fluo before running setup. So I annotated a node with fluo and ran setup again. After this...
If you configure more nodes than limits allow, you have to go in and terminate each instance manually. Is there something better that could be done? At a minimum, muchos...
When using an Accumulo snapshot tarball, Zetten tries and fails downloading the snapshot tarball from Apache. It should see that the current tarball exists and skip the download.
[collectd has a smart plugin](https://collectd.org/documentation/manpages/collectd.conf.5.shtml#plugin_smart) and it might be useful to monitor this info to look for failed disks (on metal clusters)
I ran the accumulo continuous ingest test on my baremetal cluster. The verify map-reduce job was not completing. It turns out that the home directory for the install user was...
Now that [ACCUMULO-2388](https://issues.apache.org/jira/browse/ACCUMULO-2388) is fixed, should be able to drop the following from accumulo-site.xml : ``` xml general.rpc.timeout 240s ```