Tools to process web server logs?

Hi,

I'm wondering how do you guys approach the processing of webserver logs, specially for heavy loaded servers. Please, correct me if I'm wrong, but I believe there's no information about it in the handbooks.

What could cause a simple query to take 5 seconds to execute?

Hello everyone,

I was having a terrible problem with MYSQL CPU use hitting 100% after upgrading to 4.7. I installed the devel module to see what was happening, and saw that some simple queries were taking a very long time to execute.

For example, according to the Devel module, the following query was taking 1.5 seconds (many others were taking even longer):

How to kill a perfectly good webserver

Yesterday we launched a new design for BlufftonToday.com. It all went quite well at first, and the users loved it. But about half an hour later the server began to slow to a crawl. Looking into the process table with "top" showed a big stack of running httpd processes, some of them quite old.

Restarting the webserver cleared it up, but only temporarily.

The redesign introduced several new features, including heavy use of the buddylist module, views and private messaging. Perhaps one of the modules was at fault. Or perhaps interaction with external servers (ad servers, the Spotted photo galleries, etc.) was it. Or perhaps we were under yet another botnet attack (that's become increasingly problematic lately). So we began methodically turning things on and off in a bughunt. We installed and activated APC (which had no significant effect).

After hours of stress AgentRickard suggested that we might have a 404 problem, something that dragged down Drupal when it was installed at SavannahNow.com.

It turned out not to be that, but was something very similar: Drupal was serving the directory index for the /files directory and all its children. And that was interacting with an almost insignificant bit of HTML on the front-page template, and a host of visiting web spiders, to bring down the entire site.

Spiders often examine all the link and image URLs in an HTML document and probe the underlying directories in an attempt to discover poorly linked content. If a spider sees "files/pictures/foo.jpg," it may check files/ and files/pictures. In our case Drupal was responding to those probes by serving the homepage.

Need Stats on Drupal's Scalability

We need some stats on Drupal's ability to scale for an implementation of a very large Open Source developer community portal. We're looking for any kind of stats related to the scalability of Drupal in terms of concurrent users, open connections, database queries or just about anything concrete. We want to know how others were able to scale drupal sucessfully.

Anybody successfully use na_arbitrator, OG, and TAC together yet?

Does anybody have information on how I can use na_arbitrator, OG, and TAC in a Drupal 4.7 installation?

My main reason is to hide certain taxonomic categories from certain role types when posting but still allowing them to view items posted with those tags.

database backups and maintenance page

I'm working on a db backup script that uses mysqlhotcopy.

I'd like to put up the maintenance page, do the backup, then remove the maintenance page.

The issue I'm running in to is during the time the tables are read locked, the maintenance page won't show up.

Since the maintenance page needs the database, I'm guessing that there is some kind of contention that is slowing things down and preventing the page from being displayed. Gerhard suggested that it could be the logging trying to write to the db.

Pages

Subscribe with RSS Subscribe to RSS - Deprecated - Performance and scalability