ImprovedSystemHealthNotification

Please check the status of this specification in Launchpad before editing it. If it is Approved, contact the Assignee or another knowledgeable person before making changes.

Summary

Ubuntu should improve the system health notifications. The system has to check its status regularly and tell the user about actual or possible problems.

Rationale

When installing another Linux next to Ubuntu which will be using the same swap partition, there might be problems like a changed swap space UUID. If the system runs low on memory and doesn't have any swap activated, it gets really slow. This is just one example of situations which can be identified easily and notified to the user.

Ubuntu already implements such behavior in the gnome-volume-manager, which notifies the user if the disk space on a mounted partition is getting low.

Use cases

  • Alice is wondering why her system is so slow. She clicks through the system monitor to find the problem and finds out the system didn't mount the swap partition because of some problem.
  • Bob can't work with most of his applications because his CPU is running a process on 100% CPU time. If the system told him about this rare occurance, he could have reacted earlier.
  • Carol installed a new system service but now finds out it isn't usable. Doing further research she finds out it didn't start because of some error. She would have liked some indication that one of her defined system services didn't start correctly.

Scope

The easiest implementation is probably to identify services concerned with specific "health" aspects of the system (just like gnome-volume-manager now monitors the disk "health" = remaining space). Then, these services should be improved to notify the user about problems. As a (bad) example, "smartd" in "smartmontools" monitors the SMART status of the hard disks; This service could be improved to tell the user about problems.

Alternatively, a new framework service could be written that implements certain health aspects through plugins which should be enabled by default.

Some monitoring could be done through the new upstart system, which should already receive certain "bad" events from the system infrastructure.

Design & Implementation

  • Identify what can be monitored easily (swap space, memory, CPU, sysload, system services, ...)
  • Find a monitoring service or some service close to the monitored aspect
  • Implement the notification in a way similar to gnome-volume-manager

CategorySpec

ImprovedSystemHealthNotification (last edited 2008-08-06 16:26:20 by localhost)