This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.

Bug 22802 - Deadlock while trying to connect to Pointbase
Summary: Deadlock while trying to connect to Pointbase
Status: VERIFIED FIXED
Alias: None
Product: db
Classification: Unclassified
Component: Code (show other bugs)
Version: -FFJ-
Hardware: PC Linux
: P3 blocker (vote)
Assignee: Radko Najman
URL:
Keywords: THREAD, UI
: 32553 (view as bug list)
Depends on: 28199 32603
Blocks:
  Show dependency tree
 
Reported: 2002-04-25 16:44 UTC by Jesse Glick
Modified: 2004-03-03 23:23 UTC (History)
4 users (show)

See Also:
Issue Type: DEFECT
Exception Reporter:


Attachments
Thread dump of deadlock (13.59 KB, text/plain)
2002-04-25 16:46 UTC, Jesse Glick
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Jesse Glick 2002-04-25 16:44:53 UTC
Found in FFJEE, apparently in db module. I had
just clicked on the Pointbase server node in the
Runtime tab and selected Connect. Probably no
server was running, so it was OK for it to fail. I
may have tried to expand the Pointbase connection
node too. But the entire IDE hung. See thread
dump. IMO Children.Array.initCollection may never
block, e.g. waiting for a network connection. The
connection should be made asynchronously, and
subnodes added when they are available.
Comment 1 Jesse Glick 2002-04-25 16:46:25 UTC
Created attachment 5561 [details]
Thread dump of deadlock
Comment 2 Jesse Glick 2002-04-25 16:50:43 UTC
Sorry: Orion EE build from today's sources.

After a few minutes the IDE recovered with an error dialog:

Unable to connect, cannot establish ..... (SQL-server rejected .....)

and then all was well. But if I had not been busy filing a bug, I
would have killed the IDE with kill -9 long before this.
Comment 3 Radko Najman 2002-05-02 10:36:06 UTC
This problem can occur in some special situations (e.g. overloaded
network, db server is not running) when the JDBC driver is waiting for
response for its connect request. Usually the response is quite quick
and the connection or possible error message don't hang the IDE.
I wasn't able to reproduce Jesse's a few minutes delay. On Win2000 I
got the response till 1 second, on Linux I got the response at worst
till 3 seconds.

It is risky to fix it because the bugfix requires bigger changes in
the code and it can probably affect the rest of the module. Requesting
waiver for FFJ 4.0.
Comment 4 Pavel Flaska 2002-05-30 14:54:02 UTC
I tested this issue a little bit and have done these observation:
If there is no running server on defined port, you get error message
really quickly (as Radko wrote).

The problem is, when on defined port is anything listening. It could
be stupid socket server, which does nothing, only accept connections.
Then there is a huge timeout interval.
Comment 5 Jesse Glick 2002-05-30 17:32:48 UTC
Hmm, I definitely did not have some dummy server running. I simply
started EE after building from source and tried to use it.
Comment 6 Marek Grummich 2002-07-22 08:49:24 UTC
Set target milestone to TBD
Comment 7 Radko Najman 2002-07-31 09:46:32 UTC
Request for 3.4 waiver.
For more info see the comments above.
Comment 8 John Jullion-ceccarelli 2002-08-01 09:45:26 UTC
Docs approves waiver. Leaving Patrick's name on cc for his 
records.
Comment 9 iformanek 2002-08-01 17:51:50 UTC
I agree with the waiver
Comment 10 Radko Najman 2002-09-12 14:34:00 UTC
Request for S1S41 waiver.
For more info see the comments above.
Comment 11 Radko Najman 2002-10-23 12:17:39 UTC
Requires bigger and more complex changes in the module. Will be fixed
as a part of the task #28199.
Comment 12 Jan Chalupa 2002-10-25 11:26:59 UTC
Please don't use RESOLVED LATER, keep the bug open so that it doesn't
get out of focus.
Comment 13 Radko Najman 2003-04-02 14:16:28 UTC
*** Issue 32553 has been marked as a duplicate of this issue. ***
Comment 14 _ rkubacki 2003-04-02 14:46:39 UTC
Dup of bug that is marked 'resolve later'? Apparent UI bug that *can*
be fixed. It is meant as a way how live without third waiver in a row?

Quite simple to reproduce: use wrong url for new connection to
database and if it has long timeout you've got serious trouble in 'New
Database Connection ...' dialog when you press OK. What is even worse
you have the same problem when you try to switch to Advanced tab in
the same dialog but there is no way how to find that IDE tries to
connect to database imediately. Please remember that many users are
not big wizards so they sometimes try to use wrong parameters.
Comment 15 Radko Najman 2003-04-02 15:14:23 UTC
I think it is not hard to find that this issue depends on task #28199
which is marked as STARTED and target milestone is 4.0. It is obvious
from the comments in these issues why it was marked "resolve later".
Comment 16 Jiri Kovalsky 2003-04-03 11:39:06 UTC
If you don't plan to implement this to S1S5, why doesn't it have
S1S5_WAV keyword ?
Comment 17 Radko Najman 2003-04-03 14:43:27 UTC
Priority decreased to P3, approved by QA.
The bugfix is in progress but it is not planned for 3.5. Part of this
fix is also redesign of the connect dialog which should better inform
the user about the progress (task #32603).
Comment 18 Patrick Keegan 2003-04-03 16:23:13 UTC
Is this something that needs to be release noted (for 
either NetBeans or S1S)? If so, is this text appropriate?:

"The IDE might freeze when using the Database Explorer to 
connect to a database under some circumstances, such as 
network overload, slow response from the database server, 
or retrieval of a large amount of data."
Comment 19 Radko Najman 2003-04-04 09:38:07 UTC
In my opinion it is not necessary to release noted it. This problem is
caused by the module design so it was in the module from the
beginning. It appears very rarely and that's why it wasn't reported
for long time.
Comment 20 Martin Schovanek 2003-04-04 12:09:23 UTC
I think that it should be be release noted, taking it into account
user can avoid it. Otherwise it is unpleasant surprise.
Sorry Radko.
Comment 21 Radko Najman 2003-04-04 12:17:57 UTC
OK, I agree with the proposed text.
Comment 22 Radko Najman 2003-07-30 14:52:58 UTC
Hopefully fixed as part of #28199 changes.
Comment 23 Martin Schovanek 2003-10-06 12:54:52 UTC
VERIFIED 
Comment 24 Patrick Keegan 2004-03-03 23:23:04 UTC
removing RELNOTE keyword