Project

General

Profile

Actions

User story #3567

closed

Size of /var

Added by Dennis Cabooter over 11 years ago. Updated almost 7 years ago.

Status:
Rejected
Priority:
N/A
Assignee:
-
Category:
System integration
UX impact:
Suggestion strength:
User visibility:
Effort required:
Name check:
Fix check:
Regression:

Description

When /var is full, Rudder stops working. It's difficult to know how big /var must be on a Rudder server and/or a Rudder node. Also Rudder stops working without a warning with a full /var partition.


Related issues 1 (0 open1 closed)

Is duplicate of Rudder - Bug #11739: When we have a /var full, Rudder LDAP or Postgres can be corruptedReleasedAlexis MoussetActions
Actions #1

Updated by Vincent MEMBRÉ over 11 years ago

  • Assignee set to Dennis Cabooter

When /var is full we can't log anything anymore.

I think we need to add a warning message when /var reach 95% of its size.

This should be checked at the webapp start, and logs it as a warning.

I think we could display that warning in the webapp (maybe in the 'administration -> policy server' page)

We should also check that information regularly (every day ? ) and logs it.

To sum up, here are my 3 proposals:

Warning when > 95% /var usage :
  • Log at Rudder start
  • display inside the policy server page.
  • check every Day/hour/month and log if needed

Dennis what do you think about those ideas ? would that answer to your needs about log ?

You're right, determining the size needed by Rudder is very complicated.
The postgres is what take most part of it.
We need to add a way to estimates its growth, I'll open another ticket about that subject and reference it here

Dennis, I have other questions about what's happenning when /var is full:

  • Does the webapp is still reachable ? (naviagte through pages ? I think writing new things or access to reporting are broken)
  • Is restarting works ?
Actions #2

Updated by Vincent MEMBRÉ over 11 years ago

  • Status changed from New to Discussion
Actions #3

Updated by Vincent MEMBRÉ over 11 years ago

I create a ticket about having database disk space usage and its growth estimation here #3568

Actions #4

Updated by Dennis Cabooter over 11 years ago

Actually I faced the following problems after /var was 100% full and resized afterwards:

- PostGres had errors and did not start properly
- The web interface was unresponsive
- CFengine had problems (Had to remove /var/rudder/cfengine-community/state and /var/rudder/cfengine-community/*.db)

Actions #5

Updated by Nicolas PERRON over 11 years ago

  • Target version changed from 2.5.3 to 2.5.4
Actions #6

Updated by Nicolas PERRON over 11 years ago

  • Project changed from 24 to Rudder
  • Target version changed from 2.5.4 to 2.5.5
Actions #7

Updated by Nicolas PERRON over 11 years ago

  • Target version changed from 2.5.5 to 2.5.6
Actions #8

Updated by Nicolas PERRON over 11 years ago

  • Target version changed from 2.5.6 to Ideas (not version specific)

Dennis, what do you think about the proposition of Vincent which are:

  • Have a warning message when /var reach 95% of its size
  • Have these informations on the webapp
  • have a check every day and logged
Actions #9

Updated by Jonathan CLARKE over 11 years ago

Just to clarify, the check should not be on /var but on "the partition(s) that contain /var/rudder and /var/log/rudder".

Actions #10

Updated by Benoît PECCATTE over 9 years ago

  • Category set to System integration
Actions #11

Updated by Benoît PECCATTE over 9 years ago

  • Assignee deleted (Dennis Cabooter)
Actions #12

Updated by Benoît PECCATTE almost 7 years ago

  • Is duplicate of Bug #11739: When we have a /var full, Rudder LDAP or Postgres can be corrupted added
Actions #13

Updated by Benoît PECCATTE almost 7 years ago

  • Status changed from Discussion to Rejected

Done in #11739

Actions

Also available in: Atom PDF