tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Möbus. Manfred <>
Subject JK connector - behavior if cachesize is exceeded
Date Wed, 22 Jun 2005 11:15:59 GMT
Have set up JK (1.2.13) connector from IIS to tomcat.
When I define a worker to have cachesize=1,
and there happen to come two simultaneous requests,
the second request is aborted with status 500 Internal Server Error (as seen
in IIS log file) and no bytes are returned.

The isapi redirector log contains a warning
[warn]  ajp_get_endpoint::jk_ajp_common.c (2126): Unable to get the free
endpoint for worker  from 1 slots

While this works as desired up to the point when this warning is written to
log, I would have expected that JK connector retries such a request a few
times to see if it will find a free endpoint in the next few milliseconds.

There is a worker property named "retries", and I would have supposed that
this setting will also be used in the situation described.

Is this a bug, or do you think that the current behavior is what JK
connector should do? 
In this case I think I should set cachesize to a very high value (say 100)
since I don't want to have requests aborted when there happens to be a short
peek in concurrent requests. 

If the requests were put into a queue and retried later, I could limit the
resources (open sockets) on IIS and tomcat side to a reasonable value, and
during this peek situation some requests would last a little longer, but
this would be perfectly acceptable.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message