Disclaimer: I am not asking for help; I am sharing my experiences. Apparently irrespective of the content of /etc/fstab, recent F11-on-XO1 builds are _mounting_ /var/cache/yum as a tmpfs. What gets placed into /var/cache/yum is the "indexes" of the repositories, plus actual packages as they are being downloaded.
On my XO-1s, tmpfs "partitions" (including for /var/cache/yum) are given 110 MB. For the os5 build (and previous versions), only the "primary" sqlite data (per repository) get saved in /var/cache/yum (this file for fedora is 38 MB) -- leaving room for yum to go about its business. But for the SoaS3 20090807 build (and previous versions), often (but not always) yum fetches additional "listfile" sqlite data (per repository) into /var/cache/yum (this file for fedora is 68 MB) -- and this additional fetch (for fedora) fails (runs out of room), causing yum to terminate abruptly (leaving /var/cache/yum at 100% occupancy). I do not know how come yum on SoaS3 tries to stuff that additional huge file into /var/cache/yum, when os5 does not -- but tmpfs on my XO-1 is *definitely* not big enough. The result is that on my XO-1 I have to manually "relocate" what /var/cache/yum holds, before I can consistently use 'yum' with recent soas builds. mikus _______________________________________________ Devel mailing list [email protected] http://lists.laptop.org/listinfo/devel
