[Typo3-typo3org] Downtimes of typo3.org

[Robert Lemke]

Good morning Jan-Hendrik,

On Sunday 13 March 2005 01:20, Jan-Hendrik Heuing [NF] wrote:
> We do. The server did work, but did not send correct results, that's why we
> did not get notified... We are going to change monitoring in some way.

please specify this because for me it is too fuzzy to get an idea what is 
happening. I know that such things can happen, servers stop out of the sudden 
or some other weird problems arise.  But what I need is someone who 
proactively tells we that there has been a problem, in what the problem 
consisted and how it is going to be solved.

So please explain what you mean with the server did not send correct results 
and how you are going to change the monitoring.

> > Well, you should at least set up a monitoring software plus you (or
> > Robert)
> > should upgrade to TYPO3 3.8.0 as soon as possible (Cache-control
> > headers!).
>
> Yes, I am really looking for those, but Kaspers feedback on this is "no
> timetable", so we just wait for it. As long as I am not in charge of doing
> those things myself, I can only wait for that.

Please provide me with a shell and FTP account and I update TYPO3.org. But I 
need that account longer than only for updating because you never know when a 
problem occurs and you have to switch back.

We are using Zend Accelerator, right? Then I'd also need some way to flush its 
cache myself so I can try things out without having to call someone.

> >> And if work will be done on TER, the mysql-cluster would also not be
> >> needed at all, just some failover solution which would make it easier.

Yes you are right, perhaps we don't need a MySQL cluster at all. But a 
failover solution (like a MySQL on one of the webservers) would have been 
essential already one year ago.

> >> We already have a solution for that as well. MySQL is not yet the
> >> problem, but will be very soon if traffic doubles every
> >> less-than-a-year. The version of TER right now (and rest of typo3.org,
> >> which is not much) makes use of a whole dual-xeon. But you already know
> >> those problems. We did those analysis a half a year ago...

You are right, the TER problem is known for a very long time now and Kasper 
couldn't work on it due to lack of time and I didn't work on it because I 
didn't know enough about typo3.org. But we're on a good way now I think.

> I discussed them with Daniel, Robert and Kasper independently ages ago. As
> there was no feedback, and obviously no discussion between them, we finaly
> spend our time on other things ;)

This is all true but to be honest, it has nothing to do with some other 
problems we currently have. 

We need a proper administration and set up of the servers so they don't fail - 
too much load is a different story of course. We need proper monitoring and 
information about the incidents. 

And the rest we should figure out by putting down the criteria in the wiki.

Have a nice Sunday,
robert

-- 
Robert Lemke
Assessor - TYPO3 Association
http://association.typo3.org