From scott at mikolyski.net Wed Feb 11 02:09:08 2009 From: scott at mikolyski.net (Scott Mikolyski) Date: Tue, 10 Feb 2009 15:09:08 -1000 Subject: [Udpcast] udp-cast sending too much data Message-ID: <49922534.7060904@mikolyski.net> The past few days, I've been fighting to get udpcast to send an uncorrupted disk image to my lab computers. The hard disks are 160GB (160000000000) with six partitions (a mix of ntfs and ext2/3/swap). During the multicast, the sender reports some time outs. After a few hours and 160GB later, the progress indicator says it has run out of disk space on the receivers. Not all of the partitions exist on the cloned disks with a lot of free disk space either labelled empty or as an unknown partition. I'm using the prebuilt pxe linux kernel (2.6.27.6) with a cast-o-matic initrd file. Are the casting time outs related to the data corruption? And if the disks are identical how can I be running out of space? What can I do to get a clean copy of my disk across. I've been staying up way too long trying to figure this out. Any help is greatly appreciated. From knightsamar at gmail.com Wed Feb 11 14:06:49 2009 From: knightsamar at gmail.com (Knight Samar) Date: Wed, 11 Feb 2009 18:36:49 +0530 Subject: [Udpcast] embedding driver Message-ID: Hi, The pre-build kernel for udpcast doesn't include driver for my network card hence all my pcs only boot up and can't accept ip address. How do I including the driver's .ko file in such a way that it is available for the kernel to use ? I tried the *files executables to be added to image* option in cast-o-matic, but that doesn't work. Please help! Thanks. Samar -------------- next part -------------- An HTML attachment was scrubbed... URL: From alain at knaff.lu Wed Feb 11 16:36:24 2009 From: alain at knaff.lu (Alain Knaff) Date: Wed, 11 Feb 2009 16:36:24 +0100 Subject: [Udpcast] embedding driver In-Reply-To: References: Message-ID: <4992F078.8060600@knaff.lu> Knight Samar wrote: > Hi, > > The pre-build kernel for udpcast doesn't include driver for my network card > hence all my pcs only boot up and can't accept ip address. > > How do I including the driver's .ko file in such a way that it is available > for the kernel to use ? > I tried the *files executables to be added to image* option in cast-o-matic, > but that doesn't work. > > Please help! > > Thanks. > > Samar This is option is for including user-mode executables (scripts, utilities), no kernel-mode modules. If you want to include additional modules, you need to set up udpcast with a kernel that you compiled yourself (modules must match the exact kernel version) according to the instructions at http://udpcast.linux.lu/mkimagedoc.html#kernelCompile Regards, Alain From knightsamar at gmail.com Thu Feb 12 16:49:48 2009 From: knightsamar at gmail.com (Knight Samar) Date: Thu, 12 Feb 2009 21:19:48 +0530 Subject: [Udpcast] embedding driver In-Reply-To: <4992F9F1.25358.24D6E316@mikes.kuentos.guam.net> References: <4992F9F1.25358.24D6E316@mikes.kuentos.guam.net> Message-ID: Thanks a lot Alan and Michael! :) I will get back to you after trying out both the things. I was looking for a driver for network card RTL8168. Its not supported by the udpcast kernel. Regards, Samar On Wed, Feb 11, 2009 at 9:46 PM, Michael D. Setzer II < mikes at kuentos.guam.net> wrote: > What is the driver you are looking for? > I saw the reply from Alan, but a possible option that might work. > I have a project called g4l, which does images using local and ftp options > mainly, but does also include udpcast. If it recognizes your nic, you might > be > able to use it. I had udp-reciever setup as an option on the menu, but the > udp-sender has to be run manually, so might be more than you want to do. > > So, you can download the latest g4l version 0.29 from sourceforge, and see > if it does recognize the card. Then the issue would be are you doing a full > disk image or a partition image. You will also, have to figure how the disk > is > recognized by g4l. It uses the standard kernels, and the disk may show up > at > /dev/hda instead of /dev/sda. My systems with regular ide drives are > /dev/hda, but with Fedora the come up as /dev/sda, so it doesn't matter, > but > you need the right device to get it to work. > > Also, how many machines are you doing? If it is a limited number, a regular > image could be done, and then restore via g4l regularly. Also, which OS are > your machines. G4L has ntfsclone, which is faster than an image copy of > ntfs > partitions, but would need to know the setup. > > Hope this helps... > > > On 11 Feb 2009 at 18:36, Knight Samar wrote: > > > > > Hi, > > > > The pre-build kernel for udpcast doesn't include driver for my network > card hence all my pcs only > > boot up and can't accept ip address. > > > > How do I including the driver's .ko file in such a way that it is > available for the kernel to use ? > > I tried the files executables to be added to image option in > cast-o-matic, but that doesn't work. > > > > Please help! > > > > Thanks. > > > > Samar > > > +----------------------------------------------------------+ > Michael D. Setzer II - Computer Science Instructor > Guam Community College Computer Center > mailto:mikes at kuentos.guam.net > mailto:msetzerii at gmail.com > http://www.guam.net/home/mikes > Guam - Where America's Day Begins > +----------------------------------------------------------+ > > http://setiathome.berkeley.edu (Original) > Number of Seti Units Returned: 19,471 > Processing time: 32 years, 290 days, 12 hours, 58 minutes > (Total Hours: 287,489) > > BOINC at HOME CREDITS > SETI 7,363,780.2045 | EINSTEIN 2,314,435.1909 | ROSETTA > 778,985.9265 > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From alain at knaff.lu Thu Feb 12 17:04:46 2009 From: alain at knaff.lu (Alain Knaff) Date: Thu, 12 Feb 2009 17:04:46 +0100 Subject: [Udpcast] udp-cast sending too much data In-Reply-To: <49922534.7060904@mikolyski.net> References: <49922534.7060904@mikolyski.net> Message-ID: <4994489E.2090708@knaff.lu> Scott Mikolyski wrote: > The past few days, I've been fighting to get udpcast to send an > uncorrupted disk image to my lab computers. The hard disks are 160GB > (160000000000) with six partitions (a mix of ntfs and ext2/3/swap). > During the multicast, the sender reports some time outs. After a few > hours and 160GB later, the progress indicator says it has run out of > disk space on the receivers. Not all of the partitions exist on the > cloned disks with a lot of free disk space either labelled empty or as > an unknown partition. > > I'm using the prebuilt pxe linux kernel (2.6.27.6) with a cast-o-matic > initrd file. > > Are the casting time outs related to the data corruption? And if the > disks are identical how can I be running out of space? What can I do to > get a clean copy of my disk across. I've been staying up way too long > trying to figure this out. Any help is greatly appreciated. > _______________________________________________ > Udpcast mailing list > Udpcast at udpcast.linux.lu > https://udpcast.linux.lu/cgi-bin/mailman/listinfo/udpcast You're probably running into the seek bug present in some Linux kernels. The problem is being fixed (Linux patches take a really long time to get through the chain into a released kernel...), so in the meantime I added a workaround for it to udpcast. You can download this version from the udpcast site as usual. Alain From zouxy2008 at gmail.com Sat Feb 14 14:16:00 2009 From: zouxy2008 at gmail.com (xy zou) Date: Sat, 14 Feb 2009 21:16:00 +0800 Subject: [Udpcast] compile udpcdialog failed Message-ID: I want to compile udpcast in fedora9(gcc version 4.3), my process as follows: 1) download and complile ncurses-5.7 2) sudo make install /* Install ncurses library */ 3) tar xfjv busybox-1.13.2.tar.bz2 4) cd busybox-1.13.2 5) tar xfzv ../udpcast-20081213.tar.gz 6) (cd udpcast-20081213 && ./configure) 7) tar xfzv ../udpbusybox-20081130.tar.gz 8) ../addBbApp.pl udpcdialog udpcast-20081213 9) make *defconfig* /* Create the maximum "sane" configuration. */ 10) vi Makefile, modify line to LDLIBS := pthread ncurses 11) make But i got following error message: udpcdialog/lib.a(udpc_dialog.o):(__libc_atexit+0x0): undefined reference to '_IO_cleanup' collect2: ld returned 1 exit status make: *** [busybox_unstripped] Error 1 I think that the problem is with shared libc library. nm result as follows: $ nm /usr/lib/libc.a | grep _IO_cleanup U _IO_cleanup 00001330 T _IO_cleanup 00000000 r __elf_set__libc_atexit_element__IO_cleanup__ U _IO_cleanup I need help !!! wanyozxy -------------- next part -------------- An HTML attachment was scrubbed... URL: From zouxy2008 at gmail.com Sat Feb 14 14:52:14 2009 From: zouxy2008 at gmail.com (xy zou) Date: Sat, 14 Feb 2009 21:52:14 +0800 Subject: [Udpcast] compile udpcast failed Message-ID: I list symbol from udpcdialog/lib.a 1) cd busybox-1.13.2/udpcdialog 2) nm lib.a | grep "_IO_cleanup" get result: U _IO_cleanup 00000000 T __bring_in_IO_cleanup 00000000 d __elf_set___libc_atexit_element__IO_cleanup__ _IO_cleanup not defined in busybox-1.13.2/udpcdialog/lib.a why ??? -------------- next part -------------- An HTML attachment was scrubbed... URL: From alain at knaff.lu Sat Feb 14 16:47:40 2009 From: alain at knaff.lu (Alain Knaff) Date: Sat, 14 Feb 2009 16:47:40 +0100 Subject: [Udpcast] compile udpcast failed In-Reply-To: References: Message-ID: <4996E79C.3020408@knaff.lu> xy zou wrote: > I list symbol from udpcdialog/lib.a > > 1) cd busybox-1.13.2/udpcdialog > 2) nm lib.a | grep "_IO_cleanup" > > get result: > > U _IO_cleanup > 00000000 T __bring_in_IO_cleanup > 00000000 d __elf_set___libc_atexit_element__IO_cleanup__ > > _IO_cleanup not defined in busybox-1.13.2/udpcdialog/lib.a why ??? The purpose of __bring_in_IO_cleanup is to force the linker to include the function _IO_cleanup , which is responsible for flushing IO buffers before exit. Because of static linking the linker "forgets" to include this, unless such tricks are used. Without it, some data that application write right before exit might get lost. Regards, Alain From alain at knaff.lu Mon Feb 16 09:06:31 2009 From: alain at knaff.lu (Alain Knaff) Date: Mon, 16 Feb 2009 09:06:31 +0100 Subject: [Udpcast] compile udpcdialog failed In-Reply-To: References: Message-ID: <49991E87.7020506@knaff.lu> xy zou wrote: > I want to compile udpcast in fedora9(gcc version 4.3), my process as > follows: Sorry, I only noticed this mail now, for some reason it went into my junk folder .... > > 1) download and complile ncurses-5.7 > 2) sudo make install /* Install ncurses library */ > 3) tar xfjv busybox-1.13.2.tar.bz2 > 4) cd busybox-1.13.2 > 5) tar xfzv ../udpcast-20081213.tar.gz > 6) (cd udpcast-20081213 && ./configure) > 7) tar xfzv ../udpbusybox-20081130.tar.gz > 8) ../addBbApp.pl udpcdialog udpcast-20081213 > 9) make *defconfig* /* Create the maximum "sane" configuration. */ ^^^^^^^^^^^^^^^^^ ... and I think here lies the problem. The busybox config shipped with udpcast is carefully tuned for usage with udpcast. There are to configs. The "mini" config is intended for cases where space is premium. It makes a busybox with the least number of features for udpcast to work, but in so little space that it fits on a floppy. The "maxi" config is intended for cases where space is not an issue. It makes a busybox with a number of goodies, in addition to those strictly needed for udpcast. If you need some particular feature, I'd suggest starting with the "maxi" config, and if your feature is still missing, just add what you need. > 10) vi Makefile, modify line to LDLIBS := pthread ncurses > 11) make > > But i got following error message: > udpcdialog/lib.a(udpc_dialog.o):(__libc_atexit+0x0): undefined > reference to '_IO_cleanup' > collect2: ld returned 1 exit status > make: *** [busybox_unstripped] Error 1 > > I think that the problem is with shared libc library. ^^^^^^ Exactly. In normal circumstances, with udpcast, you should compile busybox _statically_ But on the other hand, the issue that IO_cleanup is meant to fix only occurs with static compilation, so with dynamic compilation, you can indeed safely remove it. You might still run into other issue though, due to the changed busybox config. > nm result as follows: > $ nm /usr/lib/libc.a | grep _IO_cleanup > > U _IO_cleanup > 00001330 T _IO_cleanup > 00000000 r __elf_set__libc_atexit_element__IO_cleanup__ > U _IO_cleanup > > I need help !!! > > wanyozxy Regards, Alain From tomislav.parcina at email.t-com.hr Fri Feb 27 10:01:12 2009 From: tomislav.parcina at email.t-com.hr (=?ISO-8859-2?Q?Tomislav_Par=E8ina?=) Date: Fri, 27 Feb 2009 10:01:12 +0100 Subject: [Udpcast] Sending problem Message-ID: <49A7ABD8.5070201@email.t-com.hr> I have problems restoring image with Clonezilla which uses udpcast. Since it seems that it's problem in udpcast I'm sending description of the problem here. Whole discussion you can find here https://sourceforge.net/mailarchive/forum.php?thread_name=49A3C8C9.1080603%40email.t-com.hr&forum_name=clonezilla-user or here http://tinyurl.com/bk5m9p And here is explanation of problem and messages that I receive on CLI and in log files. I create image of MS Win XP and I save it to server. Then I boot clients with PXE and they start Clonezilla. After that Clonezilla server should start sending multicast (or brodcast, whichever I choose in configuration) data. 3:14 minutes after that I get this message on CLI **** apt-cache search/opt/drbl/sbin/ocs-functions: line 3537: 27660 Exit 141 ( for img in $imagedir/$target_d/$img_file_prefix; do cat $img; done ) 27662 Floating point exception| $udpcast_send_cmd 2> $udpcast_stderr **** And in /var/log/messages I have **** Feb 27 08:26:37 server udpcast[27662]: New connection from 192.168.1.104 (#0) Feb 27 08:26:37 server udpcast[27662]: first connection: min wait[0] secs - max wait[0] - min clients[4] Feb 27 08:26:47 server udpcast[27662]: New connection from 192.168.1.103 (#1) Feb 27 08:27:10 server udpcast[27662]: New connection from 192.168.1.105 (#2) Feb 27 08:27:20 server udpcast[27662]: New connection from 192.168.1.106 (#3) Feb 27 08:27:22 server udpcast[27662]: min receivers[4] reached: starting Feb 27 08:27:22 server udpcast[27662]: Starting transfer: file[] pipe[] port[2232] if[eth1] participants[4] Feb 27 08:30:36 server udpcast[27662]: dropped client #0 because of timeout Feb 27 08:30:36 server udpcast[27662]: Disconnecting #0 (192.168.1.104) Feb 27 08:30:36 server udpcast[27662]: dropped client #1 because of timeout Feb 27 08:30:36 server udpcast[27662]: Disconnecting #1 (192.168.1.103) Feb 27 08:30:36 server udpcast[27662]: dropped client #2 because of timeout Feb 27 08:30:36 server udpcast[27662]: Disconnecting #2 (192.168.1.105) Feb 27 08:30:36 server udpcast[27662]: dropped client #3 because of timeout Feb 27 08:30:36 server udpcast[27662]: Disconnecting #3 (192.168.1.106) **** I have tried to restore two images and they both end the same way (3:14 min and same message). And I can restore that image with unicast, so it's not image problem. I have tried both multicast and brodcast method, and non of them works, so I guess it's not switch issue. All clients and server are connected on one switch. Please confirm me that I'm having problems with unicast. Please help me to troubleshoot this problem. Best regards, P.S. I'm using Ubuntu 8.10 server and udpcast version is 20081213 -- Tomislav Parčina tomislav.parcina at email.t-com.hr Google talk: parcina80 From diego.c.zimmermann at gmail.com Fri Feb 27 13:00:28 2009 From: diego.c.zimmermann at gmail.com (Diego Zimmermann) Date: Fri, 27 Feb 2009 09:00:28 -0300 Subject: [Udpcast] Problems With timeout of satellite transmission Message-ID: Hi guys, I'm using udp-sender to send an announce package and using udp-receiver inside of a C program in loop who stays checking for new announce packages. The announce package has no more then 1.5K The problem is, after some interactions, udp-receiver lock and stay quiet waiting for close the connection by sender, but the announce is complete inside the system. So, my question is, can I put a timeout in udp-receiver to close this connection if some defined time pass away? -- Diego de Carvalho Zimmermann Phone: +55(41) 8836-8972 http://blogdoboot.blogspot.com "Be Free, use Linux" -------------- next part -------------- An HTML attachment was scrubbed... URL: From alain at knaff.lu Fri Feb 27 16:40:55 2009 From: alain at knaff.lu (Alain Knaff) Date: Fri, 27 Feb 2009 16:40:55 +0100 Subject: [Udpcast] Problems With timeout of satellite transmission In-Reply-To: References: Message-ID: <49A80987.60406@knaff.lu> Diego Zimmermann wrote: > Hi guys, > I'm using udp-sender to send an announce package and using udp-receiver > inside of a C program in loop who stays checking for new announce packages. > The announce package has no more then 1.5K > The problem is, after some interactions, udp-receiver lock and stay quiet > waiting for close the connection by sender, but the announce is complete > inside the system. > So, my question is, can I put a timeout in udp-receiver to close this > connection if some defined time pass away? Sorry, but in my opinion, this would not be a valid solution. Rather, try to find out why it happens in the first place. Tcpdump may help with the analysis. Which command line options do you use? Does the problem also happen on a LAN (with the same command-line parameters used). Does it only happen with small files? Regards, Alain From alain at knaff.lu Fri Feb 27 16:43:06 2009 From: alain at knaff.lu (Alain Knaff) Date: Fri, 27 Feb 2009 16:43:06 +0100 Subject: [Udpcast] Sending problem In-Reply-To: <49A7ABD8.5070201@email.t-com.hr> References: <49A7ABD8.5070201@email.t-com.hr> Message-ID: <49A80A0A.3020905@knaff.lu> Tomislav Parčina wrote: > I have problems restoring image with Clonezilla which uses udpcast. > Since it seems that it's problem in udpcast I'm sending description of > the problem here. > > Whole discussion you can find here > https://sourceforge.net/mailarchive/forum.php?thread_name=49A3C8C9.1080603%40email.t-com.hr&forum_name=clonezilla-user > or here > http://tinyurl.com/bk5m9p Does the problem also happen with udpcast alone? Does it still happen with current versions of udpcast? Regards, Alain From diego.c.zimmermann at gmail.com Fri Feb 27 17:27:02 2009 From: diego.c.zimmermann at gmail.com (Diego Zimmermann) Date: Fri, 27 Feb 2009 13:27:02 -0300 Subject: [Udpcast] Problems With timeout of satellite transmission In-Reply-To: <49A80987.60406@knaff.lu> References: <49A80987.60406@knaff.lu> Message-ID: Thanks by your replay Alain. I use udp-sender/receiver in a LAN too but the problem doesn't exists. I'm using tcpdump to check what packages it's coming by satellite, but in there looks normal. I don't know if a solution with timeout can solve my problem, but its my first choose to fix this problem. Anyone have this problem related or know anyone who have this before? Thanks a lot On Fri, Feb 27, 2009 at 12:40 PM, Alain Knaff wrote: > Diego Zimmermann wrote: > > Hi guys, > > I'm using udp-sender to send an announce package and using udp-receiver > > inside of a C program in loop who stays checking for new announce > packages. > > The announce package has no more then 1.5K > > The problem is, after some interactions, udp-receiver lock and stay quiet > > waiting for close the connection by sender, but the announce is complete > > inside the system. > > So, my question is, can I put a timeout in udp-receiver to close this > > connection if some defined time pass away? > > Sorry, but in my opinion, this would not be a valid solution. Rather, try > to find out why it happens in the first place. Tcpdump may help with the > analysis. > > Which command line options do you use? Does the problem also happen on a > LAN (with the same command-line parameters used). Does it only happen with > small files? > > Regards, > > Alain > -- Diego de Carvalho Zimmermann Phone: +55(41) 8836-8972 http://blogdoboot.blogspot.com "Be Free, use Linux" -------------- next part -------------- An HTML attachment was scrubbed... URL: From alain at knaff.lu Fri Feb 27 19:09:38 2009 From: alain at knaff.lu (Alain Knaff) Date: Fri, 27 Feb 2009 19:09:38 +0100 Subject: [Udpcast] Problems With timeout of satellite transmission In-Reply-To: References: <49A80987.60406@knaff.lu> Message-ID: <49A82C62.4050700@knaff.lu> Diego Zimmermann wrote: > Thanks by your replay Alain. > I use udp-sender/receiver in a LAN too but the problem doesn't exists. > I'm using tcpdump to check what packages it's coming by satellite, but in > there looks normal. > I don't know if a solution with timeout can solve my problem, but its my > first choose to fix this problem. > Anyone have this problem related or know anyone who have this before? > > Thanks a lot Maybe somebody might be able to help if you provided more details (such as the command-line parameters that you used)? Regards, Alain > > On Fri, Feb 27, 2009 at 12:40 PM, Alain Knaff wrote: > >> Diego Zimmermann wrote: >>> Hi guys, >>> I'm using udp-sender to send an announce package and using udp-receiver >>> inside of a C program in loop who stays checking for new announce >> packages. >>> The announce package has no more then 1.5K >>> The problem is, after some interactions, udp-receiver lock and stay quiet >>> waiting for close the connection by sender, but the announce is complete >>> inside the system. >>> So, my question is, can I put a timeout in udp-receiver to close this >>> connection if some defined time pass away? >> Sorry, but in my opinion, this would not be a valid solution. Rather, try >> to find out why it happens in the first place. Tcpdump may help with the >> analysis. >> >> Which command line options do you use? Does the problem also happen on a >> LAN (with the same command-line parameters used). Does it only happen with >> small files? >> >> Regards, >> >> Alain >> > > > From diego.c.zimmermann at gmail.com Fri Feb 27 19:51:12 2009 From: diego.c.zimmermann at gmail.com (Diego Zimmermann) Date: Fri, 27 Feb 2009 15:51:12 -0300 Subject: [Udpcast] Problems With timeout of satellite transmission In-Reply-To: <49A82C62.4050700@knaff.lu> References: <49A80987.60406@knaff.lu> <49A82C62.4050700@knaff.lu> Message-ID: Sorry, I forgott in the first mail I'm using udp-sender --async --mcast-data-address 229.1.1.1 --mcast-rdv-addr 229.1.1.1 --autostart 1 --fec 8x8/128 --max-bitrate 500k --interface eth1 --portbase 4002 -f announce and udp-receiver --nosync --mcast-rdv-addr 229.1.1.1 --interface dvb0_0 --portbase 4002 -f announce On Fri, Feb 27, 2009 at 3:09 PM, Alain Knaff wrote: > Diego Zimmermann wrote: > > Thanks by your replay Alain. > > I use udp-sender/receiver in a LAN too but the problem doesn't exists. > > I'm using tcpdump to check what packages it's coming by satellite, but in > > there looks normal. > > I don't know if a solution with timeout can solve my problem, but its my > > first choose to fix this problem. > > Anyone have this problem related or know anyone who have this before? > > > > Thanks a lot > > Maybe somebody might be able to help if you provided more details (such > as the command-line parameters that you used)? > > Regards, > > Alain > > > > > > On Fri, Feb 27, 2009 at 12:40 PM, Alain Knaff wrote: > > > >> Diego Zimmermann wrote: > >>> Hi guys, > >>> I'm using udp-sender to send an announce package and using udp-receiver > >>> inside of a C program in loop who stays checking for new announce > >> packages. > >>> The announce package has no more then 1.5K > >>> The problem is, after some interactions, udp-receiver lock and stay > quiet > >>> waiting for close the connection by sender, but the announce is > complete > >>> inside the system. > >>> So, my question is, can I put a timeout in udp-receiver to close this > >>> connection if some defined time pass away? > >> Sorry, but in my opinion, this would not be a valid solution. Rather, > try > >> to find out why it happens in the first place. Tcpdump may help with the > >> analysis. > >> > >> Which command line options do you use? Does the problem also happen on a > >> LAN (with the same command-line parameters used). Does it only happen > with > >> small files? > >> > >> Regards, > >> > >> Alain > >> > > > > > > > > -- Diego de Carvalho Zimmermann Phone: +55(41) 8836-8972 http://blogdoboot.blogspot.com "Be Free, use Linux" -------------- next part -------------- An HTML attachment was scrubbed... URL: