Back-ends for Monticello?

> Howdy...
>
> Looking for recent experiences/opinions.
>
> We're using Monticello for a lot of source that gets a fair bit of
> update
> (maybe a dozen checkins a day,
> into a set of maybe 50 packages in 5 or 6 repositories.)
>
> We've been using squeak source as the backend, on a Linux box.
>
> Every few days, the 'lost it, spin out of control at 99% CPU' problem
> strikes
> the source server, and someone has to ssh to it, kick it over,
> and then e-mail everyone to check that their recent checkins made,
> reminding the recalcitrant to 'Flush Cache Versions', and so on.
>
> This causes some of the grumpy gray-bearded men
> to become exceedingly grumpy indeed.
> Especially upsetting to the old guard is the unfortunate fact that
> you may not find out right away that your checkin has gone astray.
> (It's only the next refresh in Monticello, not the checkin itself, that
> reveals the situation to the user.)
>
> Sneering has been seen.
> Unfavorable comparisons with SVN have been heard.
> Projects based on languages whose names contain J, or even C,
> have been recalled fondly.
>
> So... we're looking for a more reliable back end.
> The alternatives seem to be FTP and WebDav.
>
> Anyone have experience/advice/guidance about using
> either an ftpd, or some WebDav enablement in an Apache on Linux,
> as the server for a Monticello package repository?

> Howdy...
>
> Looking for recent experiences/opinions.
>
> We're using Monticello for a lot of source that gets a fair bit of  
> update
> (maybe a dozen checkins a day,
> into a set of maybe 50 packages in 5 or 6 repositories.)
>
> We've been using squeak source as the backend, on a Linux box.
>
> Every few days, the 'lost it, spin out of control at 99% CPU'  
> problem strikes
> the source server, and someone has to ssh to it, kick it over,
> and then e-mail everyone to check that their recent checkins made,
> reminding the recalcitrant to 'Flush Cache Versions', and so on.

> On Apr 7, 2007, at 11:06 AM, brad fowlow wrote:
>
>> Howdy...
>>
>> Looking for recent experiences/opinions.
>>
>> We're using Monticello for a lot of source that gets a fair bit of  
>> update
>> (maybe a dozen checkins a day,
>> into a set of maybe 50 packages in 5 or 6 repositories.)
>>
>> We've been using squeak source as the backend, on a Linux box.
>>
>> Every few days, the 'lost it, spin out of control at 99% CPU'  
>> problem strikes
>> the source server, and someone has to ssh to it, kick it over,
>> and then e-mail everyone to check that their recent checkins made,
>> reminding the recalcitrant to 'Flush Cache Versions', and so on.
>
>
> Gee that would be like the Sophie Project.
> Gee one of our grumpy gray-bearded men guys offered one Million  
> Euro to have someone fix it.
> Gee there where no takers.
> Gee he didn't have one million euros anyway.
>
> Gee the Unix Socket code has some sort of problem that needs  
> fixing. There are some fixes proposed
> to accept processing queuing problems, but  they've never made the  
> main source tree.

>
> On Apr 7, 2007, at 11:21 AM, Philippe Marschall wrote:
>
> > WebDav (actually HTTP, but anyway) + Apache + Linux
> > no problems at all
>
> Yes, Apache is relatively easy to configure as a Monticello server,
> and very stable in my experience.
>
> > If you want to get fancy you can even use SSL via cUrl but this not
> > seen that much testing but it doesn't use HTTPSocket.
>
> I use it all the time. Works great.

> Bert Freudenberg wrote:
>> Gee, at least we have an intern who can wiggle the mouse which  
>> makes the server responsive again without loosing anything :)
>
> Not in my experience. If the server sits there with 99% CPU  
> consumption there is usually no way to get in there with VNC and  
> wiggle anything. It also seems to be directly related to commits -  
> usually the lock up occurs right after a check-in; probably while  
> SS is trying to write whatever it writes to disk. And usually you  
> loose that last check-in.

>From: Bert Freudenberg <[hidden email]>
>Reply-To: The general-purpose Squeak developers
>list<[hidden email]>
>To: The general-purpose Squeak developers
>list<[hidden email]>
>Subject: Re: Back-ends for Monticello?
>Date: Sat, 7 Apr 2007 23:41:33 +0200
>
>>Gee the Unix Socket code has some sort of problem that needs  fixing.
>>There are some fixes proposed
>>to accept processing queuing problems, but  they've never made the  main
>>source tree.