Bug 1481336 - satellite-installer with --noop option will fail if we follow the exact steps
Summary: satellite-installer with --noop option will fail if we follow the exact steps
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: Red Hat Satellite
Classification: Red Hat
Component: Docs Install Guide
Version: 6.2.10
Hardware: All
OS: Linux
urgent
urgent
Target Milestone: Unspecified
Assignee: Stephen Wadeley
QA Contact: Michaela Slaninkova
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2017-08-14 15:51 UTC by Christian Marineau
Modified: 2019-09-26 15:38 UTC (History)
3 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2017-08-17 15:28:01 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)

Description Christian Marineau 2017-08-14 15:51:30 UTC
Document URL: 
https://access.redhat.com/documentation/en-us/red_hat_satellite/6.2/html/installation_guide

Section Number and Name: 
7.1. UPDATING SATELLITE SERVER
     6. Perform the update

[https://access.redhat.com/documentation/en-us/red_hat_satellite/6.2/html/installation_guide/updating_satellite_server_capsule_server_and_content_hosts#updating_satellite_server_to_next_minor_version]

Describe the issue: 
If we follow the instructions step by step and the "yum update" did not update the kernel, we might not have reboot the server and the katello-service will be stopped. 

  When running the following:
  # satellite-installer --scenario satellite --upgrade --noop
  
  If will fail with postgres and httpd error because services are not responding:
/Stage[main]/Foreman::Database::Postgresql/Postgresql::Server::Db[foreman]/Postgresql::Server::Role[foreman]/Postgresql_psql[CREATE ROLE foreman ENCRYPTED PASSWORD ****]: Could not evaluate: Error evaluating 'unless' clause, returned pid 97580 exit 2: 'psql: could not connect to server: No such file or directory
 /Stage[main]/Candlepin::Database::Postgresql/Postgresql::Server::Db[candlepin]/Postgresql::Server::Role[candlepin]/Postgresql_psql[CREATE ROLE candlepin ENCRYPTED PASSWORD ****]: Could not evaluate: Error evaluating 'unless' clause, returned pid 97584 exit 2: 'psql: could not connect to server: No such file or directory
 Report processor failed: Could not send report to Foreman at https://mysatellite.com/api/reports: Connection refused - connect(2)


Suggestions for improvement: 
Option 1. Modify the documentation so that we make the user reboot no matter what.
Option 2. Add an extra step to make the user start katello-service if they want to run the --noop
Option 3. Modify the --noop option to Warn that services are stopped and they must be started.

Additional info:
This could be leading to a situation where we think that would be an issue with the upgrade and prevent the user from moving forward.

Comment 1 Stephen Wadeley 2017-08-15 08:35:42 UTC
Hello


Thank you for raising this bug.

This issue of when to reboot if a kernel update occurs has also been mentioned in relation to the principle that Satellite should not be running when `yum update` is done as having new packages present before the Satellite update script is run is "not good".


As to the issue of the Satellite being down when using the "--noop" option, see the extra steps, substeps after step 19, that were added in "6.2. Upgrading a Connected Satellite Server"[1]


[1] https://access.redhat.com/documentation/en-us/red_hat_satellite/6.2/html/installation_guide/upgrading_satellite_server_and_capsule_server#upgrading_a_connected_satellite_server

Comment 5 Stephen Wadeley 2017-08-16 14:57:24 UTC
Hello Christian

Unfortunately, it is not safe to have Satellite Server running after doing a `yum update` and before running the upgrade step.


we have this text in some sections:

"Because the above "no operation" command does not actually create the files, and some Puppet resources in the module expect them to be there, some failure messages are to be expected. "

see "Upgrading a Connected Satellite Server" for example.


Note You need to log in before you can comment on or make changes to this bug.