Bijan Parsia wrote:

> Someone already pointed out that there was such a notice yesterday. But, be
> that as it may,  Minnow doesn't actually have all that much *scheduled*
> downtime. When it goes down, tyically it's because of a crash. This has
> been made *much* more common by the addition of upload capacity to the
> Swiki, though it seems to be a more generic problem with heavy loads. I've
> spent some time trying to replicate the problem and track down the bug(s),
> but haven't had a *lot* of luck...yet.
> 

I am currently running Swikis on 3 different Linux servers. I have one lightly loaded Swiki (http://icca-chicago.org) that crashes about four times a day. To cope with this until I have a chance to debug it, I have set up a wrapper script that launches Squeak and then grabs a web page from the Swiki about once a minute. If it can't retrieve the page within 30 seconds, it kills Squeak, and repeats the cycle.  So the server is currently unavailable for < 10 minutes a day. We'll see how it behaves under more load shortly. Hopefully, we'll debug the problem before it becomes too unstable.

> Mark and I have discussed some alternative mechanisms to keep access, at
> least, (if not editing) up even when Squeak crashes. Either way, I hope
> that we can get the situation stabilized soon.
> 
> (Note also that there was some thought that it might be a hardware problem,
> but, afaict, that doesn't seem to be the case.)

I've wondered the same thing about my setup, because my other two Swikis don't really exhibit these symptoms.

> 
> FWIW, my current thought is that there's something funky going on with a
> process conflict. But it's *really* hard to tell since we're not even clear
> on the conditions under which there's a crash.

Since my Swiki crashes quite regularly, I may have some additional data on the failure conditions soon.

> 
> A lot of this discussion takes place on the PWS list, if you're interested.
> 
> Cheers,
> Bijan.