After installing and migrating all images on a rather large content site, we get out of memory errors and slowed performance when accessing `admin/content`.

Looking at XHProf (for successful runs of the page...it seems to fail particularly just after we run the transition from local to S3 files on our staging server), the following two methods stick out as consuming a lot of memory:

run_init::Resources/s3-2006-03-01.php (110 megs)
load::Resources/s3-2006-03-01.php (80 megs)

Any idea why these two S3 methods would be requiring so much memory for just the content listing HTML page (this is not the viewing of the images itself)? We have metadata caching enabled. Is there anything else we can do to optimize S3 integration?

Comments

ethanw created an issue.

ethanw’s picture

ethanw CreditAttribution: ethanw commented

This appears to be impacting our homepage as well. It appears that the AmazonS3 module is checking for the existence of files with 100+ calls to the API before loading the page, increasing load time by 4+ seconds and memory consumption by 80 megs or so.

Are there likely places that need to be optimized?