Mantis Bugtracker
  

Viewing Issue Simple Details Jump to Notes ] View Advanced ] Issue History ] Print ]
ID Category Severity Reproducibility Date Submitted Last Update
0002715 [Resin] minor always 06-05-08 02:31 06-23-08 11:53
Reporter bmeyns View Status public  
Assigned To ferg
Priority normal Resolution fixed  
Status closed   Product Version 3.1.6
Summary 0002715: Apache 2.2.x + Resin 3.1.x: frequent 503 status error messages (Service Temporarily Unavailable)
Description We're running Apache 2.2.6 with Resin 3.1.6 configured as listener for resolving JSP's (Win32).
We continiously get 503 status error messages "Service Temporarily Unavailable" (and we're not experiencing any high load on the server).
 
Any idea if there is a known conflict between Apache 2.2.x and Resin 3.1.x?
Additional Information We've run this configuration well on other servers using Apache 2.2.8 with Resin 3.0.25 and didn't experience these problems.
Attached Files

- Relationships

- Notes
(0003202)
waiwong
06-16-08 21:05

We observe similar problems:

We have two scheduled tasks, each running for about 1.5 hours. The tasks will loop over thousands of records, but accessing basically just tens of web pages. It has other normal web access across the day.

When using Resin 3.0.x, a server with 700M reserved to resin, dual PIII 1G can run very smoothly.

When switching to 3.1.6, the server hangs many times a day. It seems that resin is not releasing the objects and Java has to perform garbage collection. When garbage collection is slow, it reports server unavailable.

The same server, when switching back to 3.1.5, is performing better. It just hangs after the two major tasks are performed.

Now we switch to another server with 2G reserved for resin 3.1.6, Quad Core Xeon Pro 2G, the server does not hang, but the memory still drops quickly.

Hope this helps.
 
(0003219)
ferg
06-23-08 11:53

server/26d2

The apache/iis timeout is specific to win32. The issue is the default read timeout on the apache/iis side, which was the same as the Resin timeout. Since the default was 65s, which is too short for this timeout.

The defaults have been updated. As a workaround, you can set the

  <load-balance-socket-timeout>600s</load-balance-socket-timeout>
 

- Issue History
Date Modified Username Field Change
06-05-08 02:31 bmeyns New Issue
06-05-08 19:20 nam Project Quercus => Resin
06-16-08 21:05 waiwong Note Added: 0003202
06-16-08 21:09 waiwong Issue Monitored: waiwong
06-23-08 11:53 ferg Note Added: 0003219
06-23-08 11:53 ferg Assigned To  => ferg
06-23-08 11:53 ferg Status new => closed
06-23-08 11:53 ferg Resolution open => fixed
06-23-08 11:53 ferg Fixed in Version  => 3.1.7


Mantis 1.0.0rc3[^]
Copyright © 2000 - 2005 Mantis Group
34 total queries executed.
29 unique queries executed.
Powered by Mantis Bugtracker