Re: [9fans] trying to repair my fossil + venti system

[mirtchov@cpsc.ucalgary.ca]

> booted the cdrom, started fossil, still getting xxx1 disk is full errors
> of course, and ... there is no fscons in /srv.
>
> What would cause that? Is fossil so unhappy it won't put an fscons out
> there or is the 'boot from cd' limited in some way?
>

fossil is _hardwired_ to srv fscons by init.
/sys/src/9/boot/local.c:247 says:

	run("/boot/fossil", "-f", partition, "-c", "srv -A fboot", "-c", "srv -p fscons", 0);

so you'll need to do a "-c 'srv -p fscons'" when you start it...

I know this because I spent a day trying to fix fossil crashing while
compiled with the 9pccpuf kernels -- poor thing was trying to srv
fscons twice and instead of telling me that it can't, it just crashed
(which seems to mirror the behaviour in your case ;)

> seems it does make some sense to have something drop the score into a safe
> file somewhere ... this was on my list but obviously I waited too long.
> Any suggested locations?

I have no idea how to determine the latest score..  seems to me (from
what I understand) any valid score that you see on the disk is a
possible root of a file system...  I don't think any timestamps are
stored with the score either...

I'd try and clean up fossil as per Nemo's suggestion.  If successful
you can do a 'snap -a' and use this score as your 'last known good
state'.

do you have one of those old dot matrix printers?  clog-ing to
something like that is a very good way of having non-digital copies of
the venti score :)

[Offtopic] It's also good as an instant notification of possible
problems with the system -- once upon a time we had our firewall
software write to a dot matrix printer connected on LPT1, so as soon
as it started printing stuff (making noise!) we knew we're being
portscanned...  'twas fun -- no threat level colorizations or some
such tomfoolery, just the facts ;)

andrey