Ah, you didn't mention that T.msg is actually a JSP file. That could
make a world of difference.
The WatchDog is just telling you that a call is not returning (that
is, the thread is stuck). Therefore, my guess is that there is
something in your T.msg that is causing it to hang. It's unlikely
that this is JSPWiki's fault ;-)
/Janne
On 30 May 2008, at 21:34, Terry Steichen wrote:
It started happening again. I noticed one additional item - in the
middle of *each* of those 30 second WatchDog (80) log message
clusters,
is a single log message like this:
2008-05-30 11:16:59,879 [WatchDog for 'TC'] INFO
com.ecyrd.jspwiki.util.WatchDog - Watchable 'TP-Processor84'
exceeded timeout in state 'Filtering for URL /
LoginProc.jsp' by
579 seconds
This message appears to come from WikiJSPFilter.java [line 82].
Now, it so happens that when a user logs in, Login.jsp is called
but it
then redirects the user to LoginProc.jsp (a custom module). In turn,
LoginProc.jsp calls T.msg to check if the incoming user has any unread
messages.
Can *anyone* give me a clue as to what's going on here? Any idea
of how
to work around whatever is causing this? Any idea of what I might
look
for in T.msg (which is actually a JSP) that could possibly be
triggering
this?
This is *really* frustrating because I have no clue where to even
start
in debugging this.
PS: I tried deleting T.msg and then logging in around it. When I
called
it directly, I got the expected message that the module could not be
found. But the WatchDog messages kept coming.
On Fri, 2008-05-30 at 08:56 -0400, Terry Steichen wrote:
I'm using JSPWiki 2.6.2 and recently I've noticed a strange
problem that
just started when I began using this version. During the day,
when I'm
working on it, the system works just fine.
However, when I log into the system in the morning (or presumably,
any
time after a comparable interval), the system appears to hang. I've
discovered this is caused by the fact that my Login module passes
control to a T.msg page (to check for alerts).
When I check the log, I find I'm getting clusters of about 80 log
messages, recurring every 30 seconds. The format for each message is
this:
2008-05-29 03:42:21,702 [WatchDog for 'TC'] INFO
com.ecyrd.jspwiki.util.WatchDog - Watchable 'TP-Processor3'
exceeded timeout in state 'Generating VIEW response for
WikiPage
[TC:T.msg,ver=2,mod=Sat May 24 06:49:51 PDT 2008]' by 34707
seconds
The WatchDog messages only apply to *one particular page* -
'T.msg' (out
of thousands on the site). If I replace the version of 'T.msg'
with one
that I've used without incident for months (with JSPWiki 2.5.19), the
same thing happens. If I delete T.msg altogether, the problem
disappears.
If I stop and restart Tomcat (version 5.5.17) also disappears (for a
while).
I have no idea of what's going on. Anyone have any thoughts?
PS: When the system appears to hang, I can stop my browser and
move to
another page on the site and the system works fine - unless I
invoke the
'T.msg' page, which causes it to hang again.
