problem with amd
Dominik,
I am hoping to get a large number of debian 1.1 systems running in our
teaching laboratories next year. We use amd at our site. I have
noticed a serious problem with amd as distributed. The code enabled by
defining NEW_TOPLVL_READDIR is not well thought out. It does not take
into
account the size of the clients buffer when making up the list of auto
mount points. At our site where there are many top level auto-mount
points this causes amd to crash whenever a listing is made of a top
level auto-mount point. As updatedb is run every night this means that
the machines are unusable every morning.
We have a decision to make between using FreeBSD and Linux. The
FreeBSDers main point is that Linux networking is not as robust as that
provided by FreeBSD. It seems that all distributed versions of the
unnofficial patch of amd must have this problem even though in the
Changelog file Erik Zadok states that the code is quickly hacked
together and not tested.
I have found that Debian 1.1 networking is otherwise very reliable.
Nigel Williams (Programmer)
Queen Mary and Westfield College
Mile End Road
London E1 4NS
Telephone: 0171-975-5250
Reply to: