This e-build installs an /etc/init.d/afs startup script that 1. checks *only* for lines in /proc/mounts that contain 'vice' -- this is a dangerous assumption; people may have more than one partition that mounts to /usr/vice 2. checks *only* to see if the filesystem type 'ext2' is present in the corresponding line in /proc/mounts. If not, the script returns 1 and it fails. Reproducible: Always Steps to Reproduce: 1. Install openafs-1.2.10-r1 2. Make the cache partition filesystem ext3 and mount it as normal OR mount more than one partition to a mountpoint below /usr/vice, placing it before the cache partition in /etc/fstab 3. Reboot or, at the prompt, type '/etc/init.d/afs start' Actual Results: The following message is displayed at boot time or when the script is executed from the prompt with '/etc/init.d/afs start': >>> PLEASE CREATE A EXT2 (no reiserfs) PARTITION (of aprox. 200M) >>> AND MOUNT IT TO /USR/VICE/CACHE !!! Expected Results: 1. More properly evaluate the cache partition by checking to make sure that the line grepped from /etc/fstab contains *both* 'vice' and 'cache' 2. Allow ext3 partitions to act as cache partitions (which is safe).
Line 1 from "Expected Results" should read: 1. More properly evaluate the cache partition by checking to make sure that the line grepped from /proc/mounts contains both 'vice' *and* 'cache'
I have commited a modified ebuild (bug #59624) that fixes this and various other problems. Nothing has happend in the portage tree though...
OpenAFS needs a developer to take up maintenance.
Fixed in the latest (testing) ebuilds. The check with /proc/mounts etc. has been removed. (Amongst the reasons: with -memcache there is no need for a cache, the cache doesn't need to be on a seperate partition - let alone an ext2 partition, the cache size was being stored in two seperate places, ... It is now left to the user's responsibilities to configure it correctly.
