[PA-598] Automate the puppet-agent Solaris pipeline Created: 2016/10/06  Updated: 2017/03/10  Resolved: 2017/02/09

Status: Closed
Project: Puppet Agent
Component/s: Build Automation
Affects Version/s: None
Fix Version/s: puppet-agent 1.9.2

Type: Improvement Priority: Normal
Reporter: Kurt Wall Assignee: Unassigned
Resolution: Fixed Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
is blocked by BKR-679 Investigate potential solaris11-speci... Closed
is blocked by PA-672 Automate LDOM resets on Solaris platf... Closed
relates to IMAGES-407 Refactor Hardware_Infrastructure task... Resolved
relates to IMAGES-487 Update Hardware Infrastructure with S... Resolved
Epic Link: Automate testing for all supported puppet-agent platforms
Team: Agent
Story Points: 2
Sprint: AP 2016-11-30, AP 2016-12-14, AP 2017-01-11, AP 2017-01-25
QA Risk Assessment: No Action
QA Risk Assessment Reason: N/A [mass update]: Various build, documentation, infra, or test-related issues


Add missing components for automation:

  • Remote access to controller box to execute rollback script
  • Jenkins job to run the rollback script
  • Splice Jenkins job for running the rollback script into the existing pipelines for SPARC

Comment by James Wagner [ 2016/11/03 ]

Notes from today's discussion:

Solaris 10:

  • We need a way to reset this. Kurt Wall will get this work scheduled. Heath has prior art available. We should try to move to a single solution to reset the various types of hardware in test.

Solaris 11:

  • BKR-679 scheduled for this sprint (11/16)
  • We may want to try adding more memory to the Jenkins runner
  • Will give Kevin this current sprint to try to fix this.
  • Stan has not seen this in last 2 runs.
  • If we're unable to resolve the Beaker issue, we will go with John's suggestion to add more memory to the Jenkins runner and hope that does the trick.
Comment by Kurt Wall [ 2016/11/04 ]

In the interests of completeness, Solaris 11 will also need a reset method. We have a script in place that does this for both 10 and 11:

# /root/rollback.sh
set -e
set -x
echo "Stopping Solaris ${1} LDOM..."
ldm stop --force sol"${1}"
echo "Unbinding Solaris ${1} LDOM..."
ldm unbind sol"${1}"
echo "Rolling back Solaris ${1} LDOM..."
zfs rollback tank01/sol"${1}".bootdisk@20151207
echo "Binding Solaris ${1} LDOM..."
ldm bind sol"${1}"
echo "Starting Solaris ${1} LDOM..."
ldm start sol"${1}"

Invoke it as ./rollback.sh 10 or ./rollback.sh 11 to restart the Solaris 10 and 11 LDOMs, respectively.

Comment by Kurt Wall [ 2016/11/17 ]

http://jenkins-compose.delivery.puppetlabs.net/job/solaris_sparc_ldom_reset/ – I can demo this during the meeting.

Comment by Kenn Hussey [ 2017/01/13 ]

Kurt Wall will this be complete in time for the 1.9.0 release or should we retarget this issue for 2.0.0?

Comment by Kurt Wall [ 2017/01/13 ]

Kenn Hussey I'd say 2.0.0 is more realistic at this point.

Generated at Sun Sep 27 16:33:57 PDT 2020 using Jira 8.5.2#805002-sha1:a66f9354b9e12ac788984e5d84669c903a370049.