Initial attempt to install RH Linux 7.0 required, per the running tally kept during package selection, 685 MB. After beginning the actual installation and after the partitions had been reformatted, the installer said that /usr needed to be 23MB larger. At that time it was 776.5 MB, so apparently 800 MB was required. Backed up to package selection and pared my requests down to 613 MB and installation was successful. After installation checked partition sizes and usage with PartitionMagic and found that total usage was 631.7 MB, of which 568.2 MB was /usr. Assuming 3% overhead for file system descriptors, etc, 97% of 631.7 MB is 612.7 MB, indicating that the running tally kept during package selection was reasonably accurate. Given this, plus that /usr only needed 568.2 MB, not 800, installation should not have been halted for "insufficient disk space" on /usr.
Passed to QA to reproduce.
did your install require the second CD ...? In multi-CD installs, a large file (~90 Mb see RedHat/base/stage2.img) is temporarily transferred from CD to the partition with the most space available ... this impacts the size required to successfully install (even though the .img file is temporary) ...
Yes: the second CD was required, although briefly. With my current size of 613MB the "/usr" partition shows s total size of 776.5MB with 207.6MB available. The "/" partition, incidentally, shows a total size of 407.9MB with 357.0MB available. Of course, the attempt that didn't work (685MB) was a somewhat different mix (basically the same, but with some editing and development tools added), but this suggests that perhaps the "/" partition wasn't considered. Even if the additional 72MB had _all_ been in that partition, there would still have been plenty of room for a file ~90MB in size. Another thought --- could the file have been written into the swap partition? Mine was only 62.7MB, but if it were usable and had been increased by 40MB, I could have installed a great deal more into my system.
hmmm ... this sounds like a possible boundary condition with partition sizes that is not easy to reproduce (we're not seeing it here in our test, but we're not using the exact hardware or install steps either apparently) ... being able to reproduce this more generically is the key to understanding what may be going on ...
Closing due to inactivity.
Bug 68486 is another copy of this. It is still unfixed as of RH 7.3. "Closed due to inactivity" indeed. The word is incompetence -- or sloth if you like a shorter one.