Re: [9fans] remote access to fossil console

Re: [9fans] remote access to fossil console

[David Presotto]

A real man doesn't use tools, he just drops the filing on the
disk and reads the bits directly.

Re: [9fans] remote access to fossil console

[boyd, rounin]

> A real man doesn't use tools, he just drops the filing on the
> disk and reads the bits directly.

173000g<cr>

Re: [9fans] remote access to fossil console

[boyd, rounin]

yeah, well, you could do the whole deal with datakit.

remote kernel hackery -- the whole 9 yards ...

Re: [9fans] remote access to fossil console

[Dan Cross]

Axel Belinfante <Axel.Belinfante@cs.utwente.nl> writes:
>
> at the university I could use your mailing trick; I was also thinking
> about fossil use at home where I have to (want to) be kind of self-supportive
> (hence the ideas I sketched in my orevious post)
> I suppose nemo's printer approach makes sense there.
>

If you have a 9fat, I don't see why you couldn't snarf the latest score
and dump it into a file with a known name there.  /n/9fat/newestsc.ore
or something.

	- Dan C.

Re: [9fans] remote access to fossil console

[Geoff Collyer]

I did something similar a while ago but added openondemand:

	console=fossil dev=/srv/fscons openondemand=1
		gid=sys

I think this was to cope with restarting fossil and having to open a
new /srv/fscons.

Re: [9fans] remote access to fossil console

[Axel Belinfante]

at the university I could use your mailing trick; I was also thinking
about fossil use at home where I have to (want to) be kind of self-supportive
(hence the ideas I sketched in my orevious post)
I suppose nemo's printer approach makes sense there.

> Nemo prints them on paper which is a very good idea IMHO.  I use the
> logging jmk suggested and have a console srv-ed that drains to
> /sys/log/fossil:
>
> 	aux/clog /srv/fslog /sys/log/fossil &
>
>
> every morning the latest vac score is mailed to and stored on a
> university-administered machine, so I don't need to worry about
> backing the scores up.

Re: [9fans] remote access to fossil console

[mirtchov]

> I haven't kept up reading 9fans over the last months -- are there
> new ideas people are employing to capture (at least) the fingerprints
> in a 'safe' place (and possibly other (error) messages too?)

Nemo prints them on paper which is a very good idea IMHO.  I use the
logging jmk suggested and have a console srv-ed that drains to
/sys/log/fossil:

	aux/clog /srv/fslog /sys/log/fossil &


every morning the latest vac score is mailed to and stored on a
university-administered machine, so I don't need to worry about
backing the scores up.

andrey

Re: [9fans] remote access to fossil console

[Russ Cox]

> at the university I could use your mailing trick; I was also thinking
> about fossil use at home where I have to (want to) be kind of self-supportive
> (hence the ideas I sketched in my orevious post)
> I suppose nemo's printer approach makes sense there.

I don't bother logging the scores on my laptop.
The score is recorded in the fossil
super block -- I look through that if
I need to.  There should be a trivial command
to extract it.  Worst case I paw through
the venti arenas.

Russ

Re: [9fans] remote access to fossil console

[Axel Belinfante]

I think I did something like that on my fossil file server
I had running a while ago (after disk problems and a full fossil
it is in a 'under reconstruction' state for many months now).

I used the clog program to log the fossil console output;
this worked nicely, until the fossil filled up...

then every write to the fossil also produced an error message via
the fossil console that via clog tried to be logged to the already
overfull fossile ad finitum (nauseum...)

I'm wondering whether it would make sense to keep a kfs (or 9fat?)
partition for logging purposes -- at least to capture/save the
fossil/venti fingerprints.
I haven't kept up reading 9fans over the last months -- are there
new ideas people are employing to capture (at least) the fingerprints
in a 'safe' place (and possibly other (error) messages too?)

Axel.


> Here's something I did last week to get remote access to the fossil
> console using consolefs to our server called 'ehime':

[9fans] remote access to fossil console

[jmk]

Here's something I did last week to get remote access to the fossil
console using consolefs to our server called 'ehime':

1) you need a consoledb file, say /lib/ndb/consoledb-ehime:
	% cat consoledb-ehime
	group=sys
		uid=jmk
		uid=...
	console=ehime-fs dev=/srv/fscons
		gid=sys
	%

2) add a case to /bin/C to select the correct server:

	...
	switch($1){
	case ehime-fs
		server=ehime
	case *
		# look for server in /lib/ndb
		server=`{ndb/query sys $1 console}
		switch($server){
		case ''
			echo C: unknown console server for $1
			exit 1
		}
	}
	...

3) add startup code for consolefs in /bin/cpurc.ehime:

	...
	# start the remote fs console server
	if(test -e /srv/fscons)
		aux/consolefs -c /lib/ndb/consoledb.ehime
	...

Now we can 'C ehime-fs' just like we can for the serial consoles
of the servers in the machine room.

I changed consolefs last week not to present the status and ctl files
for remote consoles if the underlying device doesn't support them, but
this will work without that update to consolefs.