For the past few months, Apache has been dying occasionally for no obvious reason. I couldn't find out why so all I could really do was restart it. Today I caught it in the process and discovered that it had a very large number of sessions running Gnews2Rss (the famous Google news to RSS scraper) that were sitting waiting to write data out.

What seems to be happening is that many RSS aggregators do a feed collection at the top of the hour. People using gnews2rss were hitting my server, which was then kicking off a page request to Google. These were getting stacked up and taking too long to time out. Apache was then complaining about too many sessions.

So sadly I'm retiring Gnews2rss as it looks like I can no longer support the growing success. For the next few days all it will return is a single item each day repeating this message. After a week or so I'll switch that to a 410 Gone http message for an RSS request but leave the html display.

I'll leave the source code up there. And as I still use a local copy myself I'll be trying to keep the source code up to date and working.

As always, please hassle Google on mailto:news-feedback@google.com and ask them to produce RSS/Atom themselves.


[ << RSS aggregator collection timing ] [ F*ck Big Media: Rolling Your Own Network >> ]
[ 02-Nov-04 10:23am ] [ , ]