Hi, Actually, yes, I do have separate root partition which is quite small, it was a home partition with ~300GB free. Setting SQLITE_TMPDIR to home directory helped to clone it successfully. The final file is around 15GB now. Thanks.
However, I was trying to clone repo in the system with ~36 GB free as well without any separate partitions, it also failed with out of memory error. Seems like vacuuming process is taking quite a lot of space by itself. On Tue, May 21, 2019 at 7:22 PM Joerg Sonnenberger <jo...@bec.de> wrote: > > On Sun, May 19, 2019 at 08:34:44PM +0300, Andrius V wrote: > > Hello, > > > > Is there anybody successfully using fossil repository of NetBSD source > > (https://src.fossil.netbsd.org)? Some time ago I started to get errors > > on fetching which I though there were my environment issues, so I just > > switched to git repo at that time. Recently I decided to try fossil > > repo again but I am failing even to clone it. It gets "stuck" at 75.3% > > of rebuilding repository meta-data, at this point fossil file starts > > to grow indefinitely till disk size limit is reached (I had over 300 > > GB free before cloning was started) and process fails. It makes an > > impression that something is infinitely looping around at that point. > > Could it be that fossil repository is broken or can it be potentially > > a fossil bug? Or does it really just need some huge amount of disk > > memory to rebuild this meta data? Thank you. > > Do you have a separate file system for /tmp? At least for the vacuum, it > will by default use space in that. The sqlite database currently grows > to 18GB during the initial clone, don't have numbers for the transaction > log right now. That said, my file system had less than 180GB free :) > > Joerg
