PNFS Block Layout TODO
From Linux NFS
(Difference between revisions)
| Line 3: | Line 3: | ||
== pNFS Block Layout Userspace Daemon == | == pNFS Block Layout Userspace Daemon == | ||
| - | * Give it consistent naming (process name, executable name, pidfile name, etc) | + | * Give it consistent naming (process name, executable name, pidfile name, etc) '''DONE 10/28/10''' |
* Fully understand and document what it does and why it needs to do it | * Fully understand and document what it does and why it needs to do it | ||
* Use existing methods (udev, libdevmapper, etc) to scan and create disks instead of inventing new methods | * Use existing methods (udev, libdevmapper, etc) to scan and create disks instead of inventing new methods | ||
| - | * read /sys/block instead of supplying a default config file | + | * read /sys/block instead of supplying a default config file '''DONE 10/28/10''' |
* Use udev notifications to trigger device rescan, instead of rescanning once a minute | * Use udev notifications to trigger device rescan, instead of rescanning once a minute | ||
* Use a pre-existing upcall mechanism instead of inventing our own | * Use a pre-existing upcall mechanism instead of inventing our own | ||
Latest revision as of 18:12, 28 October 2010
This page is intended to serve as a "to-do" list for the Linux pNFS Block Layout client.
pNFS Block Layout Userspace Daemon
- Give it consistent naming (process name, executable name, pidfile name, etc) DONE 10/28/10
- Fully understand and document what it does and why it needs to do it
- Use existing methods (udev, libdevmapper, etc) to scan and create disks instead of inventing new methods
- read /sys/block instead of supplying a default config file DONE 10/28/10
- Use udev notifications to trigger device rescan, instead of rescanning once a minute
- Use a pre-existing upcall mechanism instead of inventing our own
- Give it meaningful debug messages
- General code cleanup
- man page and any other needed documentation
—Eanderle 18:13, 3 September 2010 (UTC)
