Bug #3397: psql error when running rudder-init.sh - Rudder - Issue Tracker

Actions

Copy link

Bug #3397

closed

psql error when running rudder-init.sh

Added by Michael Gliwinski over 11 years ago. Updated over 11 years ago.

Status:

Released

Priority:

Assignee:

Vincent MEMBRÉ

Category:

System techniques

Target version:

2.4.5

Pull Request:

https://github.com/Normation/rudder-t...

Severity:

UX impact:

User visibility:

Effort required:

Priority:

Name check:

Fix check:

Regression:

Description

Nothing critical, just a bit confusing.

When running rudder-init.sh to setup a server (2.5), and specifically when this script runs cf-agent, errors from psql are visible in the output:

R: @@DistributePolicy@@log_info@@root-DP@@root-distributePolicy@@00@@reports@@None@@2013-04-06 19:30:47+01:00##root@#slapd has been restarted
psql: FATAL:  password authentication failed for user "rudder" 
FATAL:  password authentication failed for user "rudder" 
password retrieved from file "/root/.pgpass" 
R: @@DistributePolicy@@result_repaired@@root-DP@@root-distributePolicy@@00@@reports@@None@@2013-04-06 19:30:47+01:00##root@#The Rudder Webapp configuration files were updated with a new SQL password
R: @@DistributePolicy@@result_repaired@@root-DP@@root-distributePolicy@@00@@reports@@None@@2013-04-06 19:30:47+01:00##root@#The Rudder PostgreSQL user account's password has been changed

This is a bit confusing as it appears there is nothing fatal about this error :) In fact it looks like output from psql command used in distributePolicy/1.0/passwordCheck.cf in bundle root_password_check_psql when defining class psql_cant_connect (I was able to reproduce the same errors when the password in .pgpass file is incorrect).

The command uses --output /dev/null option, but psql prints the error to stderr anyway.

Actions

Copy link

Updated by Vincent MEMBRÉ over 11 years ago

Category set to System techniques
Target version set to 2.4.5

If we can't connect to the psql database we enforce a new password to make Rudder use its database.

The two result repaired show that there had been a problem, but that the agent was able to fix it.

Maybe the message should should be clearer about the error?

I don't think we want an error report here, because we can fix it immediatly, which is the meaning of result_repaired.

I thinks this happening too in 2.4 server

Actions

Copy link

Updated by Jonathan CLARKE over 11 years ago

Status changed from New to 8
Assignee set to Vincent MEMBRÉ
Priority changed from N/A to 3

Vincent MEMBRÉ wrote:

If we can't connect to the psql database we enforce a new password to make Rudder use its database.

The two result repaired show that there had been a problem, but that the agent was able to fix it.

Maybe the message should should be clearer about the error?

Vincent, Michael's point is not that the "repaired" messages should not be displayed, but that the errors straight from PSQL should not be. Specifically these lines:

psql: FATAL:  password authentication failed for user "rudder" 
FATAL:  password authentication failed for user "rudder" 
password retrieved from file "/root/.pgpass"

At least I think so - Michael, please correct me if I'm wrong.

If that is the case, we just need to redirect stderr to /dev/null, since we analyze error based on the return codehere. Can you implement the fix please?

Actions

Copy link