[Seaside] REST and Seaside

[Nevin Pratt]

David T. Lewis wrote:

>On Fri, Apr 08, 2005 at 08:58:37AM -0700, Nevin Pratt wrote:
>  
>
>>As many of you know, I've been running Bountiful Baby on Seaside for 
>>over a year now.  Well, the site has increased in popularity to the 
>>point where the traffic appears to be a serious problem for Seaside.  I 
>>have regularly been seeing memory footprints of the Seaside image up 
>>around a gigabyte.  The latest image save was over half a gig.  And 
>>occasionally I see image freezes over extended periods of time (half a 
>>minute or so), and it is probably related to a huge image footprint.
>>    
>>
>
>Hi Nevin,
>
>I would guess that those large image saves could make your server
>appear to lock up for a half minute. 
>


Well, yes, but no that is not what is causing it.

Yesterday I did an image save (to a different image file) *because* of 
the performance problems-- it wasn't the image save that *caused* the 
problems.

What is causing it is simple image bloat, with a lot of Seaside state 
hanging around.  I think it is happening during garbage collection times 
when the image is around a gigabyte large.  And, it's really not 
surprising to see, given the size of the image when it happens.

I've got the session expiry time set large-- to 6000 seconds (100 
minutes), but if I set the session expiry time lower than that, I start 
getting complaints from the user community.  And of course, such a large 
session expiry time is going to be a problem when the site starts 
getting blasted with requests.

As I mentioned before, the image size seems to go back down after 
awhile, so the sessions seem to be expiring and getting garbage 
collected (eventually), just as we would expect.

Nevin

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/seaside/attachments/20050409/1bcc3f3f/attachment.htm