Bug #3481: cf_lock.db might grows out of control on the Rudder Server - Rudder - Issue Tracker

Actions

Copy link

Bug #3481

closed

cf_lock.db might grows out of control on the Rudder Server

Added by Nicolas CHARLES over 11 years ago. Updated almost 10 years ago.

Status:

Released

Priority:

Assignee:

Matthieu CERDA

Category:

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

In http://www.rudder-project.org/redmine/issues/2607 we implemented a mechanism to check if the cf_lock.db was preventing the agent from running, but we have overlooked the server part.
We should implement the same kind of mecanism for the server

Actions

Copy link

Updated by Matthieu CERDA over 11 years ago

Status changed from New to Discussion
Assignee set to Jonathan CLARKE
Priority changed from 1 (highest) to 2

I do not understand, this modification has been done on the failsafe and the common policy, this should work on the server too...

Actions

Copy link

Updated by Matthieu CERDA over 11 years ago

Status changed from Discussion to Pending technical review
Assignee changed from Jonathan CLARKE to Matthieu CERDA
% Done changed from 0 to 100
Pull Request set to https://github.com/Normation/rudder-techniques/pull/112

PR available.

Actions

Copy link

Updated by Jonathan CLARKE over 11 years ago

Status changed from Pending technical review to Discussion

Matthieu, this PR is necessary for Rudder 2.6, but does not address the issue in this ticket.

To explain a bit better: the promise you changed looks at the last_successful_inputs_update file and if it's too old, kills cf_lock.[tc]db. However, this file is only generated if "!root_server", see https://github.com/Kegeruneku/rudder-techniques/blob/fddad39aad3c4a87f3e68da6ac3382717aa5315f/techniques/system/common/1.0/update.st#L108. That is the problem here.

I think a simple fix is to adapt the "touch" on the last_successful_inputs_update file, and always do it if we are the root_server. That way, if cf-agent does block, it will not get touched, and we'll reset the cf_lock DB file, but not otherwise. I will create a pull request for this.

Can you create another ticket for your original pull request?

Actions

Copy link

Updated by Jonathan CLARKE over 11 years ago

Status changed from Discussion to Pending technical review
Pull Request changed from https://github.com/Normation/rudder-techniques/pull/112 to https://github.com/Normation/rudder-techniques/pull/113

Jonathan CLARKE wrote: