[Sugar-devel] Adventures in the land of ENOSPC

[Martin Langhoff]

On Sat, Sep 15, 2012 at 8:03 AM, Sascha Silbe <silbe at activitycentral.com> wrote:
> Martin Langhoff <martin.langhoff at gmail.com> writes:
>
>> Manuel Kaufmann has been looking at SL#394, and looking at the bug
>> report, it struck me that it was reported backwards. I would have
>> written: "I filled up my disk and it knocked the system out, Sugar
>> would not start again, etc. Oh, btw, it was with Browse".
>
> I agree. Every single mishandling of ENOSPC is a bug in the affected
> piece of software and should be fixed. However, there's a lot of broken

In our case, 99% of the time is the datastore. And when it borks, it
messes up its indexes so it _has_ the files and the metadata, but will
show you an empty listing for the Journal.

> Similarly, the data store should take care to leave some free

The datastore needs to handle ENOSPC correctly. That is much much
deeper than what we can do activity-side.

> Sounds like SL#2317 [1], as Gonzalo already pointed out.

Indeed, and the problem is that index_updated remains on disk all the
time. I am causing ENOSPC artificially, and creating or updating
journal entries against ENOSPC and... index_updated is never removed.

BTW, to test this I am putting sugar's own logs on tmpfs, which you
can do with a file like this:

# cat /etc/statetab.d/olpc.debug
/home/olpc/.sugar/default/logs

cheers,



m
-- 
 martin.langhoff at gmail.com
 martin at laptop.org -- Software Architect - OLPC
 - ask interesting questions
 - don't get distracted with shiny stuff  - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff