Control: tags -1 + unreproducible moreinfo Hi On Mon, Jan 16, 2017 at 04:52:09PM +0100, Andreas Beckmann wrote: > Installing mysql-server ... > Reading package lists...Bus error The actual command run can be pretty much ignored as we fail generating the binary caches in this step (which is pretty much the first one) and is not specific to a command. Reading up on "Bus error" (on wikipedia [1]) I see 3 possible reasons: - "Non-existent address". Kinda possible, althrough the insistence on "physical" vs "virtual" (which would be a Segfault) is telling. Cachebuilding is complicated (as its supposed to be done quick and at the same time general enough so we can work with it later) and I can't rule out a bug there. We sometimes even fix them if found – althrough all I have seen and fixed in the last 7 years so far were segfaults. - "Unaligned access". Certainly possible, expect that I guess piuparts runs on amd64/i386 which makes this impossible as this class does not exist on them then. Also, someone would have likely noticed as our segfault causing issues tend to be addresses not rewritten on remapping/moving – that doesn't make them magically unaligned – so the unaligned access would happen in "normal" runs as well. - "Paging errors". Quoting wikipedia: "[…] or because a just-created memory-mapped file cannot be physically allocated, because the disk is full." That sounds for me like the most likely scenario actually as our binary caches are 'just-created memory-mapped files' which depending on your sources can have a non-trivial amount of size so perhaps that is really a run out of free space (mine are both >= 100 MB). So, with that in mind and given that even the reporter can't reproduce it the bugreport as such seems rather unactionable - tagging as such. Best regards David Kalnischkies [1] https://en.wikipedia.org/wiki/Bus_error
Attachment:
signature.asc
Description: PGP signature