launchpad-reviewers team mailing list archive

[Gavin Panella <gavin.panella@xxxxxxxxxxxxx>]

On 30 May 2011 01:41, Robert Collins <robertc@xxxxxxxxxxxxxxxxx> wrote:
[...]
>> - When starting the rabbitmq-server process, also set the current
>>  working directory to the same as the newly configured HOME
>>  directory.
>
> This was already done.
> '> -        os.environ['HOME'] = kwargs['homedir']
> '
>
> So that wasn't the problem.

That only sets HOME, not the working directory, unless I've overlooked
something. I agree that this isn't likely to make a difference
though... I was stabbing in the dark.

>
>> - I realised there's (probably) no reason to daemonize the RabbitMQ
>>  server, so I've ripped out all of that complexity, wherein may lie
>>  bugs, and replaced it with subprocess.Popen().
>
> I've no particular opinion on this. I kept the daemonize logic from U1
> on the assumption that they had a reason for it.

Okay, I'll ask around to see if there is a reason.

>
> Anyhow, this can probably land but won't fix the problem.

I wondered if there might be a disk space issue. As in, Erlang can't
create the cookie because /tmp is full. IIRC, ec2 instances have the
bulk of their disk space mounted at /home or something like that, and
not /. This seems fairly unlikely though.

After that I'm out of ideas. I've looked in the Erlang source and the
error message seems to be the same whatever the cause. At first sight
Erlang appears less readable than Perl so I might be hopelessly wrong.

-- 
https://code.launchpad.net/~allenap/launchpad/rabbit-fixture-cookie-file/+merge/62709
Your team Launchpad code reviewers is requested to review the proposed merge of lp:~allenap/launchpad/rabbit-fixture-cookie-file into lp:launchpad.