Bug 42111 - rsync hangs at end of sync
Summary: rsync hangs at end of sync
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Linux
Classification: Retired
Component: rsync
Version: 7.1
Hardware: i386
OS: Linux
medium
medium
Target Milestone: ---
Assignee: Bill Nottingham
QA Contact: Aaron Brown
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2001-05-24 11:20 UTC by Jeremy Sanders
Modified: 2014-03-17 02:20 UTC (History)
2 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2001-11-30 21:03:39 UTC
Embargoed:


Attachments (Terms of Use)

Description Jeremy Sanders 2001-05-24 11:20:27 UTC
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux 2.4.2-2 i686; en-US; rv:0.8.1+)
Gecko/20010430

Description of problem:
We mirror the root partition of a machine onto a partition using the rsync
command. The command hangs at the end of the run:

[root@xpc1 jss]# rsync -raHxv --delete / /root_backup/
building file list ... done
boot/
etc/
[....]
var/run/
var/run/utmp
var/yp/binding/
var/yp/binding/xxxxxxxx.1
var/yp/binding/xxxxxxxx.2
[hangs here]

looking to see what's running....

[root@xpc1 jss]# ps auxw|grep rsync
root     10228  1.6  2.4 13512 12576 pts/6   S    12:14   0:04 rsync -raHxv
--delete / /root_backup/
root     10229  3.7  6.2 32800 31876 pts/6   S    12:14   0:11 rsync -raHxv
--delete / /root_backup/
root     10231  0.0  6.2 32800 31884 pts/6   S    12:16   0:00 rsync -raHxv
--delete / /root_backup/

[root@xpc1 jss]# strace -p 10228
select(0, NULL, NULL, NULL, {0, 10000}) = 0 (Timeout)
gettimeofday({990703222, 596410}, NULL) = 0
wait4(10229, 0xbfffdd60, WNOHANG, NULL) = 0
gettimeofday({990703222, 596485}, NULL) = 0
gettimeofday({990703222, 596508}, NULL) = 0
select(0, NULL, NULL, NULL, {0, 20000}) = 0 (Timeout)
gettimeofday({990703222, 616448}, NULL) = 0
select(0, NULL, NULL, NULL, {0, 1000})  = 0 (Timeout)
gettimeofday({990703222, 626474}, NULL) = 0
wait4(10229, 0xbfffdd60, WNOHANG, NULL) = 0
[lots more....]

Stracing PID 10229 fixes the hang!!!!



How reproducible:
Sometimes

Steps to Reproduce:
1. Don't know - now the rsync has completed properly once, it now works
every time.
2.
3.
	

Additional info:

Comment 1 Jeremy Sanders 2001-06-13 08:54:33 UTC
Discussion on Linux-kernel, see
http://marc.theaimsgroup.com/?t=99235486900004&w=2&r=1

The problem appears rsyncing a kernel tree on a local disk to a local disk.

This in _not_ an ssh problem!

Comment 2 Jeremy Sanders 2001-06-19 09:07:44 UTC
This bug appears to be fixed with the following patch:

http://www.clari.net/~wayne/rsync-nohang.patch

Comment 3 Bill Nottingham 2001-06-19 21:05:45 UTC
Added in 2.4.6-3; thanks!

Comment 4 Andreas J. Bathe 2001-06-21 05:03:35 UTC
We also try to mirror nearly 63 GBytes from a RAID 5 SCSI system to a local IDE
harddisk, the dist is RH 6.2 with 1 GB RAM, so memory or ssh shouldn't be the
problem. I also tried rsync 2.6.4 with the ~wayne/rsync-nohang.patch applied,
but the process still hangs in an undeterministic way.

In the rsync mailing list I read this email
http://www.mail-archive.com/rsync@lists.samba.org/msg01494.html which stats that
the verbose switch can be responsible for some kind of hangs. Since I left out
the -v switch the script with rsync -Ha --delete --stats --timeout=10800
--delete-excluded --exclude-from=$excludefile /export/ /export-backup runs fine.

We think that the verbose option brings rsync into a dead lock situation when
the filelist will be very big (ours is about half a million files on a very fast
raid system. So I think you should this bug be open and not resolved.

Andreas

Comment 5 Andreas J. Bathe 2001-07-16 21:31:02 UTC
Well, after nearly 4 weeks of testing the 2.4.6 version with the
~wayne/rsync-nohang.patch applied, the hang hasn't  appeared again with or
without the -verbose option. I would like to ask you to release this rsync
version as an official errata for 6.2 and above. Thanx!

Comment 6 Simon Matter 2001-08-29 06:29:22 UTC
I have experienced rsync problems yesterday for the first time in my life. It's
not exactly on RedHat Linux but I guess it's related so just let me describe
what happend.
I modified some file trees on my main ftp server (RH 6.1). What I did exactly
was to put some directories from the SAP DB 7.3 distribution CD - which is GPL -
on my ftp server containing hundreds of sudirectories and files. I then wanted
to rsync the ftp onto my laptop (Mandrake 7.2 on ReiserFS) by mounting the ftp
volume and then doing a rsync -av --delete /mnt/nfs/unix/Linux
/home/ftp/pub/software. The command was hanging several times. I went home and
wanted to rsync over to my server which is SGI XFS RedHat 7.1 with the
RH-XFS-2.4.3 kernel and kdb enabled. After syncing some files I was dropped into
kdb.
I have right now tried the rsync from RH6.1 to MDK7.2 without the -v option and
it works!!! I will remove the verbose option from my scripts now.

Comment 7 Bill Nottingham 2001-11-30 20:52:48 UTC
Does 2.5.0 (http://people.redhat.com/notting/rsync/) work for you?

Comment 8 Bill Nottingham 2002-11-25 21:55:29 UTC
closing, no response

Comment 9 Jeremy Sanders 2002-11-26 09:38:41 UTC
Sorry - I obviously forgot to reply last year! I've been using the RH 7.3 rsync
with no problems for some time now.


Note You need to log in before you can comment on or make changes to this bug.