Thanks, Paul. Your report is useful for me. I realize this problem can happen not only my source.
I hope this is a buf in JSch library and fix it. But I can prepare for the problem, because I know it happens. Regards, (2010/01/01 15:01), Paul Michael Reilly さんは書きました: > On 12/29/2009 03:55 AM, Paul Michael Reilly wrote: >> On 12/28/2009 12:12 AM, matsuda wrote: >>> Hi, >>> >>> I have a trouble using (SSH)Exec problem. >>> It is that "channel object" sometimes fails to detect the close signal. >>> Can anyone help me? >>> >>> While program does polling a remote SSH server by "ls" command, >>> "if(channel.isClosed())" sometimes doesn't change true forever! >>> >>> According to my experience, it occurs after running some nights or >>> more than 1000 times loop. >>> >>> For debug, I add below test message, then it appears. >>> It shows exit status code changed( from -1 to 0 ), >>> but channel hasn't closed. >>> if( !channel.isClosed()&& channel.getExitStatus() == 0 ){ >>> System.out.println("before close exit-status changed!" ); >>> } >>> >>> >>> Q1. Is is a known problem? >>> >>> Q2. How should I write program to recover this problem? >> These are very good questions. I have encountered both recently myself. >> I also can attest that disconnects I know to exist were not reported >> by my running application, so either I misunderstand the semantics of >> Channel.isClosed() and/or Channel.isConnected() or in fact there is a >> bug in this part of the JSch library. > > I still do not have a good feel whether there is a bug in the > Channel/Session classes wrt disconnect and closing not being detected > and reported properly. I think there is but would be hard pressed to > clearly state what it is. > > The new application code I have written to work around this very real > problem tries to force the issue by opening a short lived ChannelExec > connection and executing a simple (date) command. The theory being that > if the Session has died (been closed or disconnected but still reports a > connected status) then the death will be revealed for sure in the > attempt to transfer bytes. Once the workaround can be shown to be > working I will attempt to find the real problem with the library while > still having a useful application. I expect it will take much effort > and considerable time. > > -pmr ---------------------------------- Yukimasa Matsuda Quatre-i Science Co., Ltd. e-mail: mats...@i4s.co.jp http://www.i4s.co.jp ---------------------------------- ------------------------------------------------------------------------------ This SF.Net email is sponsored by the Verizon Developer Community Take advantage of Verizon's best-in-class app development support A streamlined, 14 day to market process makes app distribution fast and easy Join now and get one step closer to millions of Verizon customers http://p.sf.net/sfu/verizon-dev2dev _______________________________________________ JSch-users mailing list JSch-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/jsch-users