The current openafs ebuild and initscripts have several problems, here are some: - The initscript is unable to determine the CACHESIZE if the filesystem isn't ext2, the cache directory isn't on it's own partition, or the path to the device holding the cache partition is too long. - The initscript sometime hang when shutting down the machine, making an unattended reboot quite scary. - The initscript sometime leave zombie processes arround when you restart openafs. - A reemerge overwrites some configuration files (for example CellServDB and ThisCell). To correct these problems I have modified the ebuild and initscript. To make it possible to restart the AFS client without restarting the AFS server, I have split the initscript into 2 scripts, afs-server and afs-client. Reproducible: Sometimes Steps to Reproduce:
Created attachment 36892 [details] Proposed new ebuild
Created attachment 36893 [details] Proposed new initscript for openafs client
Created attachment 36894 [details] Proposed new initscript for openafs server
OpenAFS needs a developer to take up maintenance.
New ebuilds for openafs 1.2.13 (stable) and 1.3.85 (experimental) are available for testing. According to openafs-ml, 1.3.85 is currently undergoing testing so it can become 1.4rc. The initscript doesn't try to determine the CACHESIZE anymore, also because a cache doesn't necessarily need to be on a seperate partition. Setting it correctly has become the administrator's responsibility. (Though I have read something about usage of the keyword "AUTOMATIC" inside the cacheinfo file somewhere, but I need to check this) The reboot problems with the hanging and the zombie processes should be gone. Also the configuration files aren't overwritten anymore during a re-emerge. It'd be great if this ebuild could be tested to see if it solves the problems reported.
All of the four mentioned problems seem to be fixed. See latest testing ebuild in portage. We however did not implement the solution where there are two different init scripts available. If you want to discuss this, please open a new bug report.
