----------------------------------------------------------------
BEFORE YOU POST, search the faq at <http://java.apache.org/faq/>
WHEN YOU POST, include all relevant version numbers, log files,
and configuration files. Don't make us guess your problem!!!
----------------------------------------------------------------
Ok.. I see that in jserv_ajpv12.c in ajpv12_function you immeditately check
for at least 1 byte in the sockets read buffer. If you do not see that one
byte, you assume that the jvm is hung and restart it.
Shouldn't there be a configurable timeout for ping replies? If you have
many servlets running complicated tasks, then it may happen that the main
JServ thread does not get around to placing data in the socket buffer in
time... especially on a multi-processor machine where the httpd can be
running on a different processor than the java process! mod_jserv may not
get a task-switch and will be checking the read buffer way too soon!
I would think that there should be a configurable timeout between the write
and the read so that if a user knows that it may be awhile before the java
process responds to the ping from httpd they can keep the JVM from
restarting in the middle of processing.
... or maybe wait for x number of pings no returning before deciding to
restart the JVM.
- Greg
----- Original Message -----
From: Gregory A. Krasnow <[EMAIL PROTECTED]>
To: Java Apache Users <[EMAIL PROTECTED]>
Sent: Monday, November 15, 1999 9:29 AM
Subject: JVM Reset
> ----------------------------------------------------------------
> BEFORE YOU POST, search the faq at <http://java.apache.org/faq/>
> WHEN YOU POST, include all relevant version numbers, log files,
> and configuration files. Don't make us guess your problem!!!
> ----------------------------------------------------------------
>
> We are running RedHat6.0, Blackdown JDK 1.1.7, JServ 1.1b3
> We are using JServ as the application/communication layer for a large
server
> project. The servlets we are running are doing a large amount of file and
> database IO. Currently during stress testing we are finding that at some
> point in our data processing the JServ seems to restart the JVM. (We see
> watch this happen when running top in a window). We see the following in
> our mod_jserv.log at the time of the reset.
>
> [10/11/1999 12:07:15:723] (EMERGENCY) ajp12[1]: cannot scan servlet
headers
> (500)
> [10/11/1999 12:07:15:794] (ERROR) an error returned handling request via
> protocol "ajpv12"
> [10/11/1999 12:07:16:650] (EMERGENCY) ajp12: ping: no reply (-1)
> Please make sure that
> the wrapper.classpath is pointing to the correct version of
> ApacheJServ.jar
>
>
> Does anyone out there know of any reason why the JVM might reset? At
first
> we thought it might be running out of memory, so we placed a
> -mx64m on in wrapper.bin.parameter line but we never notice the java
process
> exceeding 20 or so megs.
>
> - Greg
>
> Gregory Krasnow
> Senior Software Engineer
> Stone Analytics, Inc.
> [EMAIL PROTECTED]
> Tel: 858.503.7540 x204
> Fax: 858.503.7541
>
>
>
> --
> --------------------------------------------------------------
> Please read the FAQ! <http://java.apache.org/faq/>
> To subscribe: [EMAIL PROTECTED]
> To unsubscribe: [EMAIL PROTECTED]
> Archives and Other: <http://java.apache.org/main/mail.html>
> Problems?: [EMAIL PROTECTED]
>
--
--------------------------------------------------------------
Please read the FAQ! <http://java.apache.org/faq/>
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Archives and Other: <http://java.apache.org/main/mail.html>
Problems?: [EMAIL PROTECTED]
