[Systems] runaway apache processes on sunjammer

[David Farning]

On Tue, Nov 24, 2009 at 2:52 PM, Bernie Innocenti <bernie at codewiz.org> wrote:
> On Tue, 2009-11-24 at 05:03 -0600, dfarning at sugarlabs.org wrote:
>> The overall goal is that each aslo-web node must have access the same
>> data:
>> 1. Common database
>> 2. Common activities/[files/staging] directories
>
> How does the aslo webapp actually write into these directories?
>
> If files are just being created (never moved or removed), these
> directories could be easily kept consistent by rsyncing in both
> directions every 5 minutes.
>
> If files are also being changed locally, then things get tricky. Perhaps
> unison could handle it automatically, but I'd feel safer using a
> traditional network filesystem with decent client-side caching.

The webapp use two separate layers to keep the files synchronised.
One layer is between the web-node and the other layer is between the
web nodes and and the CDN.

1. Between the web-nodes, files must be kept consistent.  When
uploading or editing a file there is no guarantee that a user will hit
that same web-node.  Thus each node must see the data (files and
database) from one click to the next.  Each web node must have full
read/write/delete access to files. A web-node can write, read, delete,
and modify individual files and directories.

2. There is more freedom between the web nodes and the CDN.  There is
a parameter in config.php to set the delay between when the staging
dir should be synced from the web-nodes to the CDN.  The CDN only
needs read access to the files

david
> --
>   // Bernie Innocenti - http://codewiz.org/
>  \X/  Sugar Labs       - http://sugarlabs.org/
>