# Notes:
# - First please read Cloudera's README.md
# - Before running it, set the ansible vars below to match your Env, in particular the script's cmdline
# - For simplicity, I decided to create a single self-contained playbook, instead of a cleaner ansible role (with the vars in the defaults etc)
# - On demand, I'm happy to refactor this into a role
# - Also: It would have been cleaner & easier to use a different repo/folder for the deployment logic,
# so we could (recursively) install the files in this repo to the target nodes (w/out ansible files)
there are 2 playbook tags (install , run ) so you can use the playbook only for the deployment (or do it in in 2 steps)
Status:
Deployed, tested and applied on our 1. customer (running CDH 6.3.x)
14.12: Deploy and apply on a customer cluster running CPD 7.1x.
Learnings:
Ensure on all nodes to have a large enough /tmp partition, IF in /opt/cloudera/parcels you have large .tar.gz (like the CDSW tar.gz that is 5+ GB in size)
OR, simply remove/rename those (too large) parcel tar.gz , before applying the upgrade
Also, in yesterdays deployment the backup of .tar.gz was not properly done (to the given backup folder), but somehow to the /tmp folder (which on 1 node ran full, so the backup was lost also)
whereas the backup of the .jar files worked as expected (to the default backup folder)
@sdevineni I could create an issue for this , but just wondering if anyone else saw this issue (or if it's somehow related to the full /tmp folder occurring)
Infos/Notes from the included comments:
install
,run
) so you can use the playbook only for the deployment (or do it in in 2 steps)Status:
Learnings: