SourceForge.net
2006-01-11 12:51:45 UTC
Bugs item #1402865, was opened at 2006-01-11 23:51
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=474136&aid=1402865&group_id=54559
Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: transport
Group: v0.7.7-alpha-3
Status: Open
Resolution: None
Priority: 5
Submitted By: Tim Anderson (tanderson)
Assigned to: Tim Anderson (tanderson)
Summary: Killed client prevents reconnect over http
Initial Comment:
If a client connected to the server over http is killed
(Control-C etc), subsequent attempts to connect block
indefinitely.
The problem arises as the server's connection pool is
trying to determine if the http connection associated
with the killed client is still valid, and this is
blocking subsequent connection attempts.
The openjms-tunnel servlet never responds to the server
as it only reads from the socket when there is a client
request - which is somewhat limited in that it assumes
clients will terminately cleanly.
There are two issues:
1. DefaultConnectionPool prevents connection matching
to occur while dead/idle connections are being
reaped.
The aim of this was to avoid dead connections from
being used, and idle connections from being reaped
prematurely.
Solution to this is to allow connection matching while
dead connections are being reaped. There is a small
chance that a dead connection will be assigned to a
live object.
2. openjms-tunnel doesn't reap dead connections.
The reaper will block indefinitely trying to
ping the tunnel connection.
Solution to this is to reap connections that haven't
been used for a period of time. This should be
configurable.
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=474136&aid=1402865&group_id=54559
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=474136&aid=1402865&group_id=54559
Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: transport
Group: v0.7.7-alpha-3
Status: Open
Resolution: None
Priority: 5
Submitted By: Tim Anderson (tanderson)
Assigned to: Tim Anderson (tanderson)
Summary: Killed client prevents reconnect over http
Initial Comment:
If a client connected to the server over http is killed
(Control-C etc), subsequent attempts to connect block
indefinitely.
The problem arises as the server's connection pool is
trying to determine if the http connection associated
with the killed client is still valid, and this is
blocking subsequent connection attempts.
The openjms-tunnel servlet never responds to the server
as it only reads from the socket when there is a client
request - which is somewhat limited in that it assumes
clients will terminately cleanly.
There are two issues:
1. DefaultConnectionPool prevents connection matching
to occur while dead/idle connections are being
reaped.
The aim of this was to avoid dead connections from
being used, and idle connections from being reaped
prematurely.
Solution to this is to allow connection matching while
dead connections are being reaped. There is a small
chance that a dead connection will be assigned to a
live object.
2. openjms-tunnel doesn't reap dead connections.
The reaper will block indefinitely trying to
ping the tunnel connection.
Solution to this is to reap connections that haven't
been used for a period of time. This should be
configurable.
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=474136&aid=1402865&group_id=54559