Nope, disabling tso and tx changed nothing.
-John
On Fri, Apr 17, 2009 at 12:24:25PM -0400, John Lockard wrote:
> On Fri, Apr 17, 2009 at 09:39:24AM +1000, James Harper wrote:
> > Does belobog have the same network adapter and kernel as your other
> > servers?
>
> It has the same ethernet (Intel e1000) as other servers and similar
> kernel, but is now a mostly legacy system. Professor has ongoing
> research which disallows me from upgrading. Unfortunately other systems
> of that same "class" don't have nearly as much data for backing up.
>
> > What brand (machine and network adapter), what kernel, and what network
> > adapter?
>
> Machine is a Dell PowerEdge 6650, Kernel is:
> 2.6.13-15.12-smp x86_64
>
> > Could be a tcp offload issue, try disabling tso and tx csum offload on
> > the fd (use ethtool)
>
> Is this correct command?
> ethtool -K eth0 tx off tso off
>
>
> > Could also be a firewall issue timing out the connection... are the sd,
> > dir, and fd on the same lan segment?
>
> The SD and DIR are the same machine. The FD is on a different LAN
> segment.
>
> > James
>
> -John
>
>
> >
> >
> > > -----Original Message-----
> > > From: John Lockard [mailto:jlockard AT umich DOT edu]
> > > Sent: Friday, 17 April 2009 01:38
> > > To: bacula-users AT lists.sourceforge DOT net
> > > Subject: [Bacula-users] Backup failing reliably repeatable
> > >
> > > Client and server at 2.4.4. Both client and server are Linux 2.6
> > >
> > > Logs from client:
> > > --------------------
> > > 15-Apr 16:53 tibor-dir JobId 2954: Start Backup JobId 2954,
> > Job=Belobog-Data-
> > > Users.2009-04-15_16.32.07.04
> > > 15-Apr 16:53 tibor-dir JobId 2954: Using Volume "100108L2" from
> > 'Scratch'
> > > pool.
> > > 15-Apr 16:53 tibor-dir JobId 2954: Using Device "NEO-LTO-1"
> > > 15-Apr 16:53 tibor-sd JobId 2954: 3307 Issuing autochanger "unload
> > slot 3,
> > > drive 0" command.
> > > 15-Apr 16:53 tibor-sd JobId 2954: 3304 Issuing autochanger "load slot
> > 21,
> > > drive 0" command.
> > > 15-Apr 16:54 tibor-sd JobId 2954: 3305 Autochanger "load slot 21,
> > drive 0",
> > > status is OK.
> > > 15-Apr 16:54 tibor-sd JobId 2954: Wrote label to prelabeled Volume
> > "100108L2"
> > > on device "NEO-LTO-1" (/dev/nst0)
> > > 15-Apr 16:54 tibor-sd JobId 2954: Spooling data ...
> > > 15-Apr 18:22 tibor-sd JobId 2954: User specified spool size reached.
> > > 15-Apr 18:22 tibor-sd JobId 2954: Writing spooled data to Volume.
> > Despooling
> > > 27,941,786,126 bytes ...
> > > 15-Apr 18:32 tibor-sd JobId 2954: Despooling elapsed time = 00:09:57,
> > Transfer
> > > rate = 46.80 M bytes/second
> > > 15-Apr 18:33 tibor-sd JobId 2954: Spooling data again ...
> > > 15-Apr 19:29 tibor-sd JobId 2954: User specified spool size reached.
> > > 15-Apr 19:29 tibor-sd JobId 2954: Writing spooled data to Volume.
> > Despooling
> > > 27,941,794,024 bytes ...
> > > 15-Apr 19:40 tibor-sd JobId 2954: Despooling elapsed time = 00:11:24,
> > Transfer
> > > rate = 40.85 M bytes/second
> > > 15-Apr 19:41 tibor-sd JobId 2954: Spooling data again ...
> > > 15-Apr 20:28 tibor-sd JobId 2954: User specified spool size reached.
> > > 15-Apr 20:28 tibor-sd JobId 2954: Writing spooled data to Volume.
> > Despooling
> > > 27,941,795,119 bytes ...
> > > 15-Apr 20:42 tibor-sd JobId 2954: Despooling elapsed time = 00:14:30,
> > Transfer
> > > rate = 32.11 M bytes/second
> > > 15-Apr 20:43 tibor-sd JobId 2954: Spooling data again ...
> > > 15-Apr 21:27 tibor-sd JobId 2954: User specified spool size reached.
> > > 15-Apr 21:27 tibor-sd JobId 2954: Writing spooled data to Volume.
> > Despooling
> > > 27,941,795,083 bytes ...
> > > 15-Apr 21:44 tibor-sd JobId 2954: Despooling elapsed time = 00:16:34,
> > Transfer
> > > rate = 28.11 M bytes/second
> > > 15-Apr 21:45 tibor-sd JobId 2954: Spooling data again ...
> > > 15-Apr 22:29 tibor-sd JobId 2954: User specified spool size reached.
> > > 15-Apr 22:29 tibor-sd JobId 2954: Writing spooled data to Volume.
> > Despooling
> > > 27,941,795,185 bytes ...
> > > 15-Apr 22:43 tibor-sd JobId 2954: Despooling elapsed time = 00:14:28,
> > Transfer
> > > rate = 32.19 M bytes/second
> > > 15-Apr 22:44 tibor-sd JobId 2954: Spooling data again ...
> > > 15-Apr 23:29 tibor-sd JobId 2954: User specified spool size reached.
> > > 15-Apr 23:29 tibor-sd JobId 2954: Writing spooled data to Volume.
> > Despooling
> > > 27,941,793,858 bytes ...
> > > 15-Apr 23:42 tibor-sd JobId 2954: Despooling elapsed time = 00:12:54,
> > Transfer
> > > rate = 36.10 M bytes/second
> > > 15-Apr 23:43 tibor-sd JobId 2954: Spooling data again ...
> > > 16-Apr 00:26 tibor-sd JobId 2954: User specified spool size reached.
> > > 16-Apr 00:26 tibor-sd JobId 2954: Writing spooled data to Volume.
> > Despooling
> > > 27,941,794,859 bytes ...
> > > 16-Apr 00:43 tibor-sd JobId 2954: Despooling elapsed time = 00:16:46,
> > Transfer
> > > rate = 27.77 M bytes/second
> > > 16-Apr 00:43 tibor-sd JobId 2954: Spooling data again ...
> > > 16-Apr 02:07 belobog-fd JobId 2954: Fatal error: backup.c:1087 Network
> > send
> > > error to SD. ERR=Connection reset by peer
> > > 16-Apr 02:07 tibor-dir JobId 2954: Error: Bacula tibor-dir 2.4.4
> > (28Dec08):
> > > 16-Apr-2009 02:07:38
> > > Build OS: x86_64-unknown-linux-gnu redhat Enterprise
> > release
> > > JobId: 2954
> > > Job: Belobog-Data-Users.2009-04-15_16.32.07.04
> > > Backup Level: Full
> > > Client: "belobog" 2.4.4 (28Dec08)
> > x86_64-unknown-linux-
> > > gnu,suse,10.0
> > > FileSet: "belobog-data-users" 2009-04-11 05:15:01
> > > Pool: "Radev-Full" (From Job FullPool override)
> > > Storage: "Overland_Neo_4000" (From Pool resource)
> > > Scheduled time: 15-Apr-2009 16:32:02
> > > Start time: 15-Apr-2009 16:53:13
> > > End time: 16-Apr-2009 02:07:38
> > > Elapsed time: 9 hours 14 mins 25 secs
> > > Priority: 30
> > > FD Files Written: 7,029,945
> > > SD Files Written: 0
> > > FD Bytes Written: 217,674,494,521 (217.6 GB)
> > > SD Bytes Written: 0 (0 B)
> > > Rate: 6543.6 KB/s
> > > Software Compression: None
> > > VSS: no
> > > Storage Encryption: no
> > > Volume name(s): 100108L2
> > > Volume Session Id: 50
> > > Volume Session Time: 1239651099
> > > Last Volume Bytes: 194,987,520,000 (194.9 GB)
> > > Non-fatal FD errors: 0
> > > SD Errors: 0
> > > FD termination status: Error
> > > SD termination status: Error
> > > Termination: *** Backup Error ***
> > > --------------------
> > >
> > > On the client side I see this in the firewall log:
> > > Apr 16 02:07:38 belobog kernel: SFW2-OUT-ERROR IN= OUT=eth0
> > SRC=xxx.xxx.xxx.32
> > > DST=xxx.xxx.yyy.25 LEN=52 TOS=0x00 PREC=0x00 TTL=64 ID=59630 DF
> > PROTO=TCP
> > > SPT=11310 DPT=9103 WINDOW=1460 RES=0x00 ACK RST URGP=0 OPT
> > > (0101080A3D08C91922123CB5)
> > >
> > > (SRC and DST have been cleansed in the firewall log by me)
> > >
> > > Looking at information on SuSEfirewall (iptables with wrappers) I'm
> > led to
> > > believe
> > > the firewall log message is just a notice. Basically that the
> > connection was
> > > dropped, and not by the firewall.
> > >
> > > I can repeat this over and over on this same machine with a backup of
> > this
> > > partition or any other partition which is at least 300G in size (I
> > don't
> > > have any partitions larger than 217G but smaller than 200G). All of
> > these
> > > backups fail at roughly 217G as the FD Bytes Written.
> > >
> > > This backup runs from the client to the backup server with a 100G
> > spool
> > > area. From the spool, the data is written to LTO2 tape. Backups on
> > this
> > > system failed at 217G when my spool was 50G, 10G and 5G in size.
> > >
> > > I have other servers with more storage on which backups run fine, up
> > to
> > > 4TB.
> > >
> > > Any ideas?
> > >
> > > Thanks,
> > > John
> > >
> > > --
> > > Brain: Pinky, Are you pondering what I'm pondering?
> > > Pinky: Well, I think so, Brain, but I can't memorize
> > > a whole opera in Yiddish.
> > > -------------------------------------------------------------------
> > > John M. Lockard | U of Michigan - School of Information
> > > Unix and Security Admin | 1214 SI North - 1075 Beal Ave.
> > > jlockard AT umich DOT edu | Ann Arbor, MI 48109-2112
> > > www.umich.edu/~jlockard | 734-615-8776 | 734-647-8045 FAX
> > > -------------------------------------------------------------------
> > >
> > >
> > ------------------------------------------------------------------------
> > ------
> > > Stay on top of everything new and different, both inside and
> > > around Java (TM) technology - register by April 22, and save
> > > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
> > > 300 plus technical and hands-on sessions. Register today.
> > > Use priority code J9JMT32. http://p.sf.net/sfu/p
> > > _______________________________________________
> > > Bacula-users mailing list
> > > Bacula-users AT lists.sourceforge DOT net
> > > https://lists.sourceforge.net/lists/listinfo/bacula-users
> >
> >
>
> --
> Brain: Pinky, Are you pondering what I'm pondering?
> Pinky: Uh, I think so Brain, but you'd have to take
> that whole bridge apart then, wouldn't you?
> -------------------------------------------------------------------
> John M. Lockard | U of Michigan - School of Information
> Unix and Security Admin | 1214 SI North - 1075 Beal Ave.
> jlockard AT umich DOT edu | Ann Arbor, MI 48109-2112
> www.umich.edu/~jlockard | 734-615-8776 | 734-647-8045 FAX
> -------------------------------------------------------------------
>
> ------------------------------------------------------------------------------
> Stay on top of everything new and different, both inside and
> around Java (TM) technology - register by April 22, and save
> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
> 300 plus technical and hands-on sessions. Register today.
> Use priority code J9JMT32. http://p.sf.net/sfu/p
> _______________________________________________
> Bacula-users mailing list
> Bacula-users AT lists.sourceforge DOT net
> https://lists.sourceforge.net/lists/listinfo/bacula-users
>
>
--
"I think the best pictures are often on the edges of any situation,
I don't find photographing the situation nearly as interesting as
photographing the edges." - William Albert Allard
-------------------------------------------------------------------
John M. Lockard | U of Michigan - School of Information
Unix and Security Admin | 1214 SI North - 1075 Beal Ave.
jlockard AT umich DOT edu | Ann Arbor, MI 48109-2112
www.umich.edu/~jlockard | 734-615-8776 | 734-647-8045 FAX
-------------------------------------------------------------------
------------------------------------------------------------------------------
Stay on top of everything new and different, both inside and
around Java (TM) technology - register by April 22, and save
$200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
300 plus technical and hands-on sessions. Register today.
Use priority code J9JMT32. http://p.sf.net/sfu/p
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users
|
