dbbaskette/cape

Cluster Automation for Pivotal Education

cape's Issues

Ability to delete a cluster of nodes

Should be able to delete a cluster of nodes.

Add swap to adhere to system requirements

Should add a 50Gb swap to follow system requirements

uncompress cannot find file

[2016-12-13 14:16:06,637] /Users/jpatel/PE/cape/ClusterBuilder/InstallGPDB.py {InstallGPDB:uncompressFiles:515} DEBUG Thread-94 - [u'unzip: cannot find or open ./greenplum-clients-4.3.8.1-build-1-RHEL5-x86_64.zip, ./greenplum-clients-4.3.8.1-build-1-RHEL5-x86_64.zip.zip or ./greenplum-clients-4.3.8.1-build-1-RHEL5-x86_64.zip.ZIP.\n']

Maybe this got renamed to: greenplum-loaders-4.3.8.1-build-1-RHEL5-x86_64.zip or network.pivotal.io is returning new bits.

Users should be able to set GUCs for GPDB

When testing various releases of GPDB, you may need to set specific GUCs. This feature allows users to set specific GUCs by setting SET_GUCS=yes in their config.env file and editing ./templates/set_specific_GUCs.cape with the needed GUCs.

Implement option to create a RAID0 volume

Make it so that you can choose to have a RAID0 volume across multiple drives in your GPDB cluster.

Check required parameters for GPDB Install

Check for all the variables we need to successfully deploy a GPDB cluster. If required vars are missing, then fail and report on the missing variables.

Verify should check gpdb before doing detailed checks

If gpinit fails, verify should not try and do detailed segment checks as the master might be in utility mode and gpinit might have failed.

Ability to check cluster state

We should be able to check if cluster nodes are in running state. Basic for now.

Implement logging feature

This will help with troubleshooting and users can use a flag to set the log level

Make Standby and Access Nodes optional to install

Install Pre-Release GPDB

A user should be able to specify a pre-released version of GPDB they would like to install. This is not available on network.pivotal.io and should be provided to CAPE using an absolute path in config.env as below:

GPDB_BUILD=<path to binary to upload & install to deployed cluster>

Usually users would do this for testing a specifc feature or fixing bugs before they get released out to network.pivotal.io for public consumption.

Remove Pivotal Education Pieces from Build

Yes, the project was constructed for PivEd Clusters and yes its named for that, but it shoud first meet a more generic usecase. Then perhaps a second project will layer ontop.

Temporary fix for Key Exchange

Revert to using gpssh-exkeys for now as key exchange is still broken.