Messages in this thread | | | From | "Zachary Williams" <> | Subject | Accept() problem | Date | Wed, 17 Nov 1999 13:29:42 -0500 |
| |
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> <HTML><HEAD> <META content="text/html; charset=iso-8859-1" http-equiv=Content-Type> <META content="MSHTML 5.00.2722.1300" name=GENERATOR> <STYLE></STYLE> </HEAD> <BODY bgColor=#ffffff> <DIV><FONT face=Arial size=2>BUG: In a load balancing enviornment going through hardware such as a<BR>server-iron, apache can be flooded with requests (this could be due to a<BR>traffic spike) and will eventually stop responding. The load balancer will<BR>then remove that server from 'active' status, because it fails to respond to<BR>http health checks. In most cases, only a few of the children ever recieve<BR>requests. Because of the limited children actually responding to requests,<BR>the load-balancer will never put the server back to 'active' status,<BR>therefor leaving the server down, until its children are killed (killing the<BR>parent, and restarting apache. a -HUP WILL NOT WORK!).<BR><BR>Info: This is NOT, and I stress, is NOT an apache bug! All children are<BR>healthy, and are either locked, waiting to become the next child with<BR>accept(15, or the one is on accept(15 (If you have it setup to use the<BR>serialized accept) or there is the trampling herd if not. Either way, it is<BR>affected. The bug, appears to be with the kernel. We have been able to<BR>make this problem occur on kernel 2.2.9 and 2.2.12. 2.2.13, and 2.2.14pre6. </FONT></DIV> <DIV><FONT face=Arial size=2>We believe the bug is in the (buggy) wake-one code that was beggining to be </FONT></DIV> <DIV><FONT face=Arial size=2>included in kernel 2.2.8 or so (however, this is just speculation, I'll leave it up </FONT></DIV> <DIV><FONT face=Arial size=2>to the guru's to tackle this one. :) ). We have been UNABLE to reproduce this </FONT></DIV> <DIV><FONT face=Arial size=2>failure on the latest development kernel 2.3.28.<BR></FONT></DIV> <DIV><FONT face=Arial size=2>This is a difficult to reproduce bug, however, given a<BR>load-balancing setup, it is very obvious, because of the certain conditions<BR>met. (requests MUST stop going to the affected server, otherwise children<BR>will respawn, and act normally.) Single server users will notice a<BR>'slowdown' period, that lasts anywhere from 30 seconds to a few minutes,<BR>while the system kicks back into gear.<BR><BR>I am not sure if I'm properly subscribed to this list, so please CC <A href="mailto:admin@ztnet.com">admin@ztnet.com</A> with any responses. Thanks.</FONT></DIV> <DIV><FONT face=Arial size=2></FONT> </DIV> <DIV><FONT face=Arial size=2>Zach<BR></DIV></FONT></BODY></HTML>
| |