rsync - Jan 2005 - [Bug 2208] cygwin version of rsync sometimes hangs

https://bugzilla.samba.org/show_bug.cgi?id=2208


wayned@samba.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|major                       |normal
             Status|NEW                         |ASSIGNED
            Summary|rsync failed at random file |cygwin version of rsync
                   |with same error rsync:      |sometimes hangs
                   |writefd_unbuffered failed to|
                   |write 4 bytes: phase        |
                   |"unknown" [sender]:         |
                   |Connection reset by peer    |
                   |(104) rsync error: error in |
                   |rsync protocol data stream  |
                   |(code 12) at                |
            Summary|/home/lapo/packaging/tmp/rsy|
                   |nc-2.6.3/io.c(515)          |




------- Additional Comments From wayned@samba.org  2005-01-02 18:37 -------
The error you cite just indicates that the other end of the connection went
away. Your trace tells us that it went away because it timed out, so this looks
like it just boils down to the old issue of the cygwin version of rsync
sometimes hanging (though this is the first I've seen of someone encoutering
the
hang when using daemon mode).  I've never been able to duplicate this hang,
so
I've never been able to investigate it.

This is the first mention of this hang in bugzilla (previous discussion has been
confined to the mailing list), so this will be the official spot to help out
with the rsync-on-cygwin hanging problem.

-- 
Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

https://bugzilla.samba.org/show_bug.cgi?id=2208





------- Additional Comments From michal.mu@seznam.cz  2005-01-03 03:42 -------
Server side (other end of connection) went away more than 9 minutes later
than client, so server do not close TCP connection (it remains open until
server timeout - tested by netstat). But client reports TCP connection closed
nearly immeditely after last file reported. There is problem at client side,
correct me if I am wrong please.

This client hang can be reproduced anytime on file set I use 
(cca 500 MB of files). I am not able to finish any single rsync transfer.

At diffrent client PC which has same configuration - at least both clients PCs
was istalled from same sourcs and I do not see any diffrences (Win XP SP2, same
version of Cygwin) everything is OK and no hangs happend.

How can I help to investigate bug ?

-- 
Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

https://bugzilla.samba.org/show_bug.cgi?id=2208





------- Additional Comments From bmueller@rentmanager.com  2005-01-18 13:36
-------
I am having a very similar problem.  Possibly my comments can crack this issue.

Using several different servers.  Slackware with rsync 2.6.3 and FreeBSD 5.2.1 
with rsync 2.6.3 and clients (several) running cygwin 1.5.12(0.116/4/2) on 
various versions of windows.

My testbed systems are Windows XP Pro SP2 and a Windows 2003 Server System.  I 
am trying to backup these systems to any one of the rsync servers.

My command on the client is:

C:\rsync\rsync.exe --compress --delete --ignore-errors --recursive --verbose --
times --modify-window=5 --stats --
progress '/cygdrive/C/'
'username@host.server.com::share_name/test_server/test_
set/C/'

What seems to happen is that transfer will stop on the client at some point in 
time (say 2:00am) and then the client and the server will both get errors 
about an hour later.  The client gets

rsync: writefd_unbuffered failed to write 16385 bytes: phase "unknown"
[sender]: Connection reset by peer (104)
rsync: read error: Connection reset by peer (104)
rsync error: error in rsync protocol data stream (code 12) 
at /home/lapo/packaging/tmp/rsync-2.6.3/io.c(515)

The server gets...

rsync: writefd_unbuffered failed to write 43 bytes: phase "unknown" 
[generator]: Broken pipe (32)
rsync error: error in rsync protocol data stream (code 12) at io.c(915)

Looking at these it would seem that at some point the client and server simply 
stop communicating even though they are still connected.  Possibly a network 
hiccup happened in the middle of a transfer, didn't see anything obvious in 
minitoring the nextowrk.  They then enter this "unknown" state and
while they
are there they hang out for 1 hour or so - during this time they appear to 
think they are still connected.  After this time they both die with the above 
errors.

This is not specific to any fileset or any specific file.  It might happen 
after running for several days, or after a few minutes.

-- 
Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

https://bugzilla.samba.org/show_bug.cgi?id=2208





------- Additional Comments From blackb1rd@seat-ibiza.nl  2005-01-19 03:00
-------
Well I'm not having exactly the same problem, since my win32 clients does
rsync
well, but sometimes (not always) hangs after it's finished. So it won't
get back
to console after the additional status information has been printed.

I'm using the same rsync batch script on 8 different clients, and only at 2
of
them this problem (sometimes) occur. Unfortunately it's not reproducible on
demand.

-- 
Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

https://bugzilla.samba.org/show_bug.cgi?id=2208


bryan@arcamax.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bryan@arcamax.com




------- Additional Comments From bryan@arcamax.com  2005-01-24 20:53 -------
I too am having this problem.  I had been working fine for years.  I just
upgraded to all new hardware.  I went from Win2k to WinXP and from what ever I
had before to current version of cygwin/rsync.

I use this in my edit/compile/test cycle.  The source tree is on the WinXP box.
 I run a script to rsync(on ssh) the tree to the server and then compile with
ssh.  Since upgrading the rsync fails about 90% of the time with:
rsync error: error in rsync protocol data stream (code 12) at
/home/lapo/packaging/tmp/rsync-2.6.3/io.c(515)

The client dies after a few seconds and the server sits there for a while.  The
server is running Fedora Core 1 with rsync 2.5.7.

-- 
Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

https://bugzilla.samba.org/show_bug.cgi?id=2208





------- Additional Comments From bryan@arcamax.com  2005-01-26 05:58 -------
I tried copying rsync.exe from an older but working installation (Win2k).  This
was rsync 2.6.0.  That did not fix the problem.  Tonight I will try copying in
an entire working c:\cygwin tree and see if that helps.

I have determined that the problem only seems to occur when pushing files up to
the server.  I have NOT seen problems doing:
- retrieving new files from the server
- dry runs to see what is different
- upload runs where no files need to be sent.


-- 
Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.