Closed GoogleCodeExporter closed 9 years ago
In my case it works as intended. During the download a rsync starts and
transmitts a corrupt file - this is because Lsyncd using an asynchronous
monitoring interface cannot know if someone is still accessing the file. After
the file finished it receives a "CLOSE_WRITE" and starts another rsync and that
will overwrite-repair the corrupt trunks. See if there is a rsync process
running, and if wait a little longer?
Original comment by axk...@gmail.com
on 16 Aug 2011 at 6:52
So i have looked deeper on my systems. i see only one rsync process that is
running. after the wget is finished this file will not update this file. it's a
special configuration needet fot this "overwrite-repair" ?
my configuration (importent configurations)
settings: delay=60; statusInterval=60; maxDeplay=60
sync: default.rsyncssh; excludeFrom; rsyncOps="-l6tus"
Original comment by fragst...@gmail.com
on 16 Aug 2011 at 8:16
I just tried it again with rsyncssh, for me it works. Pleases wait until the
download is finished, the 60 time delay after it is over and no rsync are
running anymore.
If the problem persists, please make a minimal test case, provide complete
configuration and a complete logfile (-log all)
Original comment by axk...@gmail.com
on 16 Aug 2011 at 8:44
Hi,
i will send you this information directly and not over the portal.
the logfile are without corrections of server or usernames.
please delete after watching.
the timecode if the logfile will be interesting are 23:49:09
the original file has a size of: 493.879.296 and the transfered file of:
135.642.256
the tar file included the logfile and my configuration.
thx for help
Thomas
Original comment by fragst...@gmail.com
on 16 Aug 2011 at 10:13
As discussed by email, there seems to be an issue with timestamps and --update.
Whats the exact timestamps?
Please run `stat FILENAME' on source and target. If you manage also possibly
between first and second rsync call.
Or run "while true; do stat FILENAME; sleep 1; done;" on the target to get a
log how it changes.
Original comment by axk...@gmail.com
on 18 Aug 2011 at 9:43
i think it's an error of the rsync process
source
stat debian-6.0.2.1-amd64-netinst.iso
File: „debian-6.0.2.1-amd64-netinst.iso“
Size: 176160768 Blocks: 344408 IO Block: 4096 reguläre Datei
Device: 90e3h/37091d Inode: 51692368 Links: 1
Access: (0644/-rw-r--r--) Uid: ( 0/ lala) Gid: ( 0/ lala)
Access: 2011-08-18 16:21:24.000000000 +0200
Modify: 2011-06-28 14:59:06.000000000 +0200
Change: 2011-08-18 16:21:24.000000000 +0200
remote
stat debian-6.0.2.1-amd64-netinst.iso
File: „debian-6.0.2.1-amd64-netinst.iso“
Size: 53728356 Blocks: 104944 IO Block: 4096 reguläre Datei
Device: 811h/2065d Inode: 58733658 Links: 1
Access: (0644/-rw-r--r--) Uid: ( 1000/lala) Gid: ( 1000/lala)
Access: 2011-08-18 16:19:25.309667468 +0200
Modify: 2011-08-18 16:18:54.000000000 +0200
Change: 2011-08-18 16:19:25.309667468 +0200
the modification times are older on the remote site as this on the source.
Original comment by fragst...@gmail.com
on 18 Aug 2011 at 2:35
Not the modify timestamp! The timestamps on the source are quite strange..
Original comment by axk...@gmail.com
on 18 Aug 2011 at 2:43
Sure! The timestamp come from the remote server where i have downloaded this
file
Original comment by fragst...@gmail.com
on 18 Aug 2011 at 3:09
So rsync didn't preserve the modification timestamp on the first sync, like -t
should tell it to?
Original comment by axk...@gmail.com
on 18 Aug 2011 at 3:13
Any news about the timestamp issue?
Original comment by axk...@gmail.com
on 26 Aug 2011 at 7:54
sorry. i have not spend time for this project but i think that the problems can
be occurd by using an ubuntu alpha version on the second site.
if i have any news i will comment this here
Original comment by fragst...@gmail.com
on 29 Aug 2011 at 5:51
so i have checkt this issue on two other systems.
the result is the same.
he start a copy and on finish the files will not resynced
the tests was made with version 2.0.5 and an centos 6 system as destination
target and an debian/ubuntu system as source.
i have used other systems as before to exclude an issue from an virtual
maschine.
Original comment by fragst...@gmail.com
on 14 Sep 2011 at 3:00
You can easily workaround the issue by removing -u "--update". Otherwise you
would need to find out why rsync creates the file with wrong timestamps in your
case.
Original comment by axk...@gmail.com
on 16 Sep 2011 at 10:14
Original issue reported on code.google.com by
fragst...@gmail.com
on 16 Aug 2011 at 6:27