Drupal + eAccelerator = Apache segfault

Posted by odnomzagi on March 5, 2006 at 3:30pm

I have a busy site with Drupal 4.6.5, Apache + eAccelerator. Unfortunately Apache segfaults quite often, but random. Only restart helps. Does anyone else experience this?

Comments

I can confirm this. I'm

Posted by steffen on March 5, 2006 at 6:07pm

I can confirm this. I'm running eAccelerator 0.9.4 as Zend extension, PHP 4.4.2 and Apache 1.3.34. Apache segfaults after complaining that Allowed memory size of 8388608 bytes exhausted (tried to allocate 808960 bytes) - the numbers are of course different every error.

I just tried setting memory_limit = 12582912 in my php.ini to see if that helps.

Yes.

Posted by Boris Mann on March 5, 2006 at 6:23pm

This is eaccelerator, not Drupal, and is a known issue. Regular restarting of Apache on a cron schedule helps.

Segmentation faults - any new solutions?

Posted by David N on May 3, 2007 at 8:15pm

I apologize for bumping an old thread, but I felt it was better to continue a discussion rather than start a whole new thread.

I've read through some old posts about EAccelerator and the seemingly inevitable segmentation faults that occur. Is this still an expected downside to using EAccelerator or any PHP accelerator? Have there been any developments that prevent this from happening, or is restarting Apache still the only way around it?

I'm just planning ahead and downtime is of course a concern. As such, I would like to have a cronjob set up before launching so that I can minimize or monitor downtime, instead of implementing EAccelerator sometime down the road and dealing with unexpected downtime then.

---
David N
http://www.autoconcourse.com (coming soon)