[TUHS] Happy birthday, core dumped

Noel Chiappa jnc at mercury.lcs.mit.edu
Thu Jun 19 07:21:38 AEST 2014


    > From: "A. P. Garcia" <a.phillip.garcia at gmail.com>

    > that's like asking george martin for his source regarding a beatles
    > song...

Reminds me of the person on Wikipedia who tried to argue with me about the
'History of the Internet' article... :-)


    > From: John Cowan <cowan at mercury.ccil.org>

    >> scj at yaccman.com scripsit:

    >> a Dec repair person who ran "preventive maintenance" on our disc that
    >> wiped out the file system! His excuse was that Dec didn't support
    >> "permanent storage" on the disc at the time...

    > Next time, mount a scratch monkey.

It was probably a fixed-head disk (RS11 or RS04); can't exactly stick a
different pack in! :-) Probably the DEC OS's only used it for swapping or
something, since they were both relatively small - 512KB.

(Speaking of RS11's: the first PDP-11 I used - an 11/20 running RSTS - had a
grand total disk storage of _one_ RS11!)


And speaking of putting file systems on them: I recently wrote this command
for V6 called 'si' which allowed me (among many other interesting things) to
watch the contents of the disk buffer(s). It turns out that even with other
packs mounted, the buffer is almost always completely full of blocks from the
root device; it makes plain the value of having the root on a _really_
fast disk. 

I don't know if that usage pattern is because /bin is there, or because pipes
get created on the root, or what. When I get up the energy I'll move /bin to
another drive (yeah, yeah, I know - good way to lose and create a systen that
won't boot, so I'll actually make a _copy_ of /bin and mount it _over_ the
original /bin - probably a host of interesting errors there, e.g. if a process
has the old /bin as its current dir), and see what the cache contents look
like then.

	Noel



More information about the TUHS mailing list