updatedb (cron event) should not be started when a laptop is on battery! doing so kills the battery due to harddrive useage. i'm thinking to avert this issue, either /etc/init.d/crond needs to be start/stopped via acpi events (messy as this would shut down crond entirely), or, creating a /etc/init.d/updatedb (/etc/init.d/locate) service to start/stop manually... As a quick hack, i guess one could deal with stopping crond entirely when on battery. (since it may alse contain logrotate and other items not necessarily needed during battery mode) Reproducible: Always Steps to Reproduce: 1. 2. 3. Expected Results: SOB (Save Our Battery)
Interesting idea. I'm very hesitant to do things like altering services, cron, etc based on acpid by default like this. If you'd like to code something up like this which is easily configurable to be disabled and enabled, that would be great. I'd prefer not to have this by *default* though. Also stopping cron entirely for one annoying service is not really a good thing. People on laptops might be extensively using cron to schedule other important things they want run. We'd really need something more nuanced than stopping cron
Another thing that kinda bugs me are services being started, when X could be started first and these "other" start-up services could be executed to start using a lower priority (to allow X primary usage of the hdd) until the services are started, then they would resume thier normal priority. This is a side note from the stated topic of this feature request/bug, however, even crond, gpm, cupsd, fetchmail, ... could all be started after X is started for those of us using a laptop/computer as a client computer.. for access our PIM's on Evolution, Mozilla, etc. However, as I mentioned in the above paragraph, the disk usage after X is started would spike due to these misc services being started after X is started. So as to why I some how proposed that these misc services to restricted to the amount of hdd bandwidth they get to start. Just an idea. i'm going to keep kicking this one around for awhile. True, most that are using a standard computer (networked, server, whatnot) will rarely reboot and not need to worry about this, but those of us using a laptop not as a server (or even those users rarely using their computers until they turn it on) may benefit from such a scheme. This sort of reminds me of pcmcia/cardbus schemes. It's kinda difficult to mess with a scripting system once it works. -- as to why i *hacked* my orinoco card to startup on eth1, etc. so might first have to hack init or something to get a preliminary example working. Any ideas on how to restrict hdd bandwidth/pipe usage to a specific executing process would be great. Can it be even done yet?
The original request just sounds too messy to me. I don't think Gentoo should be implementing shutting down services based on battery levels, thats for the user to setup if they wish imho. Roger, you kinda went off on a tangent, so please open a fresh bug if you'd like too.. :)
