Red Hat Bugzilla – Bug 118916
scp dies with Lost Connection
Last modified: 2007-11-30 17:06:54 EST
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040116 Description of problem: After the server has been running for a couple of weeks, scp from local machine to remote server dies with Lost Connection. There was no network problems at that time. Strace of the server end shows 1590 rt_sigaction(SIGPIPE, {SIG_IGN}, NULL, 8) = 0 1590 close(6) = 0 Full Straces from client and server will be uploaded. Once it gets into this state restarting ssh doesn't fix it only a reboot seems to work. Scp from remote server to client works with no problems. Version-Release number of selected component (if applicable): openssh-3.1p1-14 How reproducible: Sometimes Steps to Reproduce: 1.wait till ssh gets into broken state 2.scp file to server 3. get sigpipe and local host reports Lost Connection Actual Results: nate@localhost nate]$ scp abuse.log nate@servername.redhat.com:/tmp nate@servername.redhat.com's password: abuse.log 5% |* | 392 KB 01:34 ETAlost connection Expected Results: Completion of scp. Additional info:
Created attachment 98753 [details] Strace from the Server side This is the strace output from the server.
Created attachment 98754 [details] client side strace of it dieing
Does it still happen? If yes, does ssh-ing to the server fail? Could you attach log produced with scp -vvv when it fails?
This problem has gone away with one of the latest updates.
OK, thanks.