Re: using rsync 3.0.0 CVS version

On Tue 30 Jan 2007, Paul Slootman wrote: > Still going strong :) Still running ;) One note: don't forget to fix the copyright notice on rsync, s/2006/2007/ : # rsync --version rsync version 3.0.0cvs protocol version 30 Copyright (C) 1996-2006 by Andrew Tridgell, Wayne Davison, and others.

Re: using rsync 3.0.0 CVS version

On Tue 30 Jan 2007, Wayne Davison wrote: > On Tue, Jan 30, 2007 at 02:38:42PM +0100, Paul Slootman wrote: > > It also did the sync in under 2 hours, whereas the non-incremental > > version would take up to 12 hours, and use up to 500MB of memory. > > Awesome! That's an even more dramatic improvem

Re: using rsync 3.0.0 CVS version

On Tue, Jan 30, 2007 at 05:21:35PM +0200, Shai wrote: > Same here. I did my initial test and where before it used to hang, it > doesn't now :) Thanks for the update! I appreciate your help in the testing. If you have any speed comparisons between old and new rsync runs for your setup (approximat

Re: using rsync 3.0.0 CVS version

On Tue, Jan 30, 2007 at 02:38:42PM +0100, Paul Slootman wrote: > It also did the sync in under 2 hours, whereas the non-incremental > version would take up to 12 hours, and use up to 500MB of memory. Awesome! That's an even more dramatic improvement than what I was anticipating for large hierarch

Re: using rsync 3.0.0 CVS version

Same here. I did my initial test and where before it used to hang, it doesn't now :) Thanks for tonight's update! Shai On 1/30/07, Paul Slootman <[EMAIL PROTECTED]> wrote: On Mon 29 Jan 2007, Wayne Davison wrote: > On Mon, Jan 29, 2007 at 06:33:19PM +0100, Paul Slootman wrote: > > Unfortunate

Re: using rsync 3.0.0 CVS version

On Mon 29 Jan 2007, Wayne Davison wrote: > On Mon, Jan 29, 2007 at 06:33:19PM +0100, Paul Slootman wrote: > > Unfortunately the current CVS version (updated a couple of hours ago) > > still hangs :( > > I found another potential hang scenario that could happen if the > generator was having to wait

Re: using rsync 3.0.0 CVS version

On Mon, Jan 29, 2007 at 06:33:19PM +0100, Paul Slootman wrote: > Unfortunately the current CVS version (updated a couple of hours ago) > still hangs :( I found another potential hang scenario that could happen if the generator was having to wait for a new file list to arrive, but failed to tell th

Re: using rsync 3.0.0 CVS version

On Mon, Jan 29, 2007 at 06:33:19PM +0100, Paul Slootman wrote: > Just now, with various straces running, the sending process got to: The important process in the generator, since it controls all the work. (It is the first process on the receiving side, and forks the receiver). Attaching to the gen

Re: using rsync 3.0.0 CVS version

On Sat 27 Jan 2007, Wayne Davison wrote: > I had not encountered this hang until today. The backtrace implicated > a problem in the wait_for_receiver() routine, and I figured out that > every now and then the io_flush() call could end up reading the last > available message from the receiver, giv

Re: using rsync 3.0.0 CVS version

Sweet! I'll test it soon! On 1/27/07, Wayne Davison <[EMAIL PROTECTED]> wrote: On Sun, Jan 21, 2007 at 08:43:56AM +0200, Shai wrote: > When I start the rsync, either with the rsync protocol or rsh, i found > that it'll start doing the rsync and just halt after a few hundred MBs > or even up to

Re: using rsync 3.0.0 CVS version

On Sun, Jan 21, 2007 at 08:43:56AM +0200, Shai wrote: > When I start the rsync, either with the rsync protocol or rsh, i found > that it'll start doing the rsync and just halt after a few hundred MBs > or even up to a couple GBs. I had not encountered this hang until today. The backtrace implicat

Re: using rsync 3.0.0 CVS version

Any updates on this hanging issue being fixed? On 1/23/07, Paul Slootman <[EMAIL PROTECTED]> wrote: On Mon 22 Jan 2007, Paul Slootman wrote: > > It's the same binary, compiled from rsync-HEAD-20070120-2211GMT. I tried again with current cvs (with the 1.194 version of receiver.c), and it still

Re: using rsync 3.0.0 CVS version

On Mon 22 Jan 2007, Paul Slootman wrote: > > It's the same binary, compiled from rsync-HEAD-20070120-2211GMT. I tried again with current cvs (with the 1.194 version of receiver.c), and it still hangs when transferring an empty directory (it is created on the receiver though). A local transfer (r

Re: using rsync 3.0.0 CVS version

On Mon 22 Jan 2007, Wayne Davison wrote: > On Mon, Jan 22, 2007 at 10:04:10PM +0100, Paul Slootman wrote: > > This was to another system, over Gbit ethernet; both sides are amd64, > > running in 64 bits mode. > > Make sure that both sides are running the exact same code for protocol > 30 (because

Re: using rsync 3.0.0 CVS version

On Mon, Jan 22, 2007 at 10:04:10PM +0100, Paul Slootman wrote: > This was to another system, over Gbit ethernet; both sides are amd64, > running in 64 bits mode. Make sure that both sides are running the exact same code for protocol 30 (because the protocol is still in flux). If one side is older

Re: using rsync 3.0.0 CVS version

On Mon 22 Jan 2007, Matt McCutchen wrote: > I can't reproduce this hang in the current rsync from CVS. I tried > pushing an empty directory to a daemon on the same machine, but rsync > finished successfully. This was to another system, over Gbit ethernet; both sides are amd64, running in 64 bits

Re: using rsync 3.0.0 CVS version

On 1/22/07, Paul Slootman <[EMAIL PROTECTED]> wrote: I just noticed that rsync-HEAD-20070120-2211GMT will hang when transferring an empty directory: I can't reproduce this hang in the current rsync from CVS. I tried pushing an empty directory to a daemon on the same machine, but rsync finished

Re: using rsync 3.0.0 CVS version

On Sun 21 Jan 2007, Wayne Davison wrote: > Another thing you can do in the debugger when attached to the generator > is to output a summary of the file-list info: > > p *first_flist > p *first_flist->next > p *first_flist->next->next I just noticed that rsync-HEAD-20070120-2211GMT will hang when

Re: using rsync 3.0.0 CVS version

On 1/21/07, Wayne Davison <[EMAIL PROTECTED]> wrote: I'm thinking that I will add a debug-output signal handler, probably listening on SIGALRM. This will let the user kill -14 any rsync 3.0.0 process to have it mention its role and some debug info. I like this idea! Matt -- To unsubscribe or

Re: using rsync 3.0.0 CVS version

On Sun, Jan 21, 2007 at 08:23:25AM -0500, Matt McCutchen wrote: > I had a similar problem, and Wayne is in the process of fixing it. > Right now he's waiting for me to provide backtraces for the hung > processes, which I will do very soon. Another thing you can do in the debugger when attached to

Re: using rsync 3.0.0 CVS version

On 1/21/07, Shai <[EMAIL PROTECTED]> wrote: I've wanted to start working with this CVS version cuz the new "incremental-recursion algorithm" is just what I need. But ran into a problem... When I start the rsync, either with the rsync protocol or rsh, i found that it'll start doing the rsync and