Re: binary diff
On Thu, Apr 3, 2008 at 11:03 AM, Alon Altman [EMAIL PROTECTED] wrote: what about cmp? thanks, i didn't know about 'cmp'. however, it's output is cryptic to me. i can not understand what the diff is On Thu, Apr 3, 2008 at 12:45 AM, Erez D [EMAIL PROTECTED] wrote: hi i am looking for something like 'diff' which can compare binary files and give a result other then just 'differ' or 'same' the same as what diff does for text files. i though of using xxd with a regular diff, but that doesn't help, as if there is one byte missing, all lines are shifted which means they are different textually, when they are not really. thanks, erez.
binary diff
hi i am looking for something like 'diff' which can compare binary files and give a result other then just 'differ' or 'same' the same as what diff does for text files. i though of using xxd with a regular diff, but that doesn't help, as if there is one byte missing, all lines are shifted which means they are different textually, when they are not really. thanks, erez.
Re: binary diff
There are a few of them, take a look for example at xdelta and bsdiff Ely On Thu, Apr 3, 2008 at 10:45 AM, Erez D [EMAIL PROTECTED] wrote: hi i am looking for something like 'diff' which can compare binary files and give a result other then just 'differ' or 'same' the same as what diff does for text files. i though of using xxd with a regular diff, but that doesn't help, as if there is one byte missing, all lines are shifted which means they are different textually, when they are not really. thanks, erez. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: binary diff
Erez - Perhaps you could first convert the binary files to text representation (od --address-radix none) and then preform the comparison in that dimension? Yaacov Erez D wrote: hi i am looking for something like 'diff' which can compare binary files and give a result other then just 'differ' or 'same' the same as what diff does for text files. i though of using xxd with a regular diff, but that doesn't help, as if there is one byte missing, all lines are shifted which means they are different textually, when they are not really. thanks, erez. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Replacing kernel 2.4 - 2.6
Muli Ben-Yehuda wrote: On Wed, Apr 02, 2008 at 11:53:36PM +0300, Shachar Shemesh wrote: As a side note, I don't know why the user space interface does not include those programs (well, I can understand about ext2 tools, assuming ext changed format since 2.4, which I don't know whether it has). We don't break compatibility lightly, but sometimes it's needed. Note that these are all system utilities which have tight interactions with the kernel. Obviously it's better if no changes are needed, but sometimes they are. I guess that's why I was referring to mount(2) and not mount(8). The first is a published and documented system call, and I will find it odd if the kernel allowed itself to break it. I don't recall exactly when or why the procps version was bumped, but you might want to take a look at /proc/meminfo sometime - it's highly kernel version specific. And yes, ps parses it. ok. I still think it shouldn't bother ps to do its main functionality, but I can see why that would matter. You have it backwards. 2.6 relies on udev to create /dev. No udev will exist on Aviram's old system, hence no /dev by default. You could probably work around it by creating a static /dev. I believe you are the one who has got it backwards. Under Redhat 7.2, dev WAS created statically. In general, many of the answers you gave here seem to me to apply to the opposite case, where an old kernel is dropped into a new userspace. My point is that the kernel (or glibc, or any other major piece of system software) usually has subtle dependencies on other pieces of system software. My point is that user space backwards compatibility is meaningless if there is no, well, user space backwards compatibility. Relying on the insides of file systems in motion (such as sysfs was at the early versions of 2.6) is sort of acceptable, but I really would have preferred it if Documentation/filesystems/proc.txt would say something along the lines of the interface is guaranteed to have the field name, a bunch of white spaces, a number and a unit (which can be kB, mB or gB only). In other words, if the docs gave all I needed to write a user space program that, if didn't actually work under all conditions, at least could tell when it doesn't have enough info to work, for all future versions. Otherwise, it seems to me, that the stable userspace pledge is just a facade. Shachar = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Replacing kernel 2.4 - 2.6
On Thu, Apr 03, 2008 at 12:19:32PM +0300, Shachar Shemesh wrote: My point is that the kernel (or glibc, or any other major piece of system software) usually has subtle dependencies on other pieces of system software. My point is that user space backwards compatibility is meaningless if there is no, well, user space backwards compatibility. Relying on the insides of file systems in motion (such as sysfs was at the early versions of 2.6) is sort of acceptable, but I really would have preferred it if Documentation/filesystems/proc.txt would say something along the lines of the interface is guaranteed to have the field name, a bunch of white spaces, a number and a unit (which can be kB, mB or gB only). In other words, if the docs gave all I needed to write a user space program that, if didn't actually work under all conditions, at least could tell when it doesn't have enough info to work, for all future versions. Otherwise, it seems to me, that the stable userspace pledge is just a facade. You are mixing two different things. The system call interface is stable and syscalls are never removed, only added. A 10-year-old binary should - in theory - run on 2.6.25-rc8. The kernel, however, employs a number of userspace helpers such as mount(8). The interface between those helpers and the kernel (e.g., format of /proc files) does change occasionally. The kernel developers however do understand that just because an interface exists for the use of a limited subset of helpers does *not* mean nothing else will use it (i.e., something else will use it), and take extra care to only change the interface when absolutely necessary. Cheers, Muli = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: binary diff
xxd's function is exactly to convert from binary into text. Try to convert your binary files using: xxd -p -c1 (works for me: xxd -v output is xxd V1.10 27oct98 by Juergen Weigert). I checked also how to do it using od. The following would work (my od --version is 5.97): od -v -w1 -ta --address-radix none or: od -v -w1 -tx1 --address-radix none --- Omer On Thu, 2008-04-03 at 11:37 +0300, Yaacov Fenster - System Engineering Troubleshooting and other stuff wrote: Erez - Perhaps you could first convert the binary files to text representation (od --address-radix none) and then preform the comparison in that dimension? Yaacov Erez D wrote: hi i am looking for something like 'diff' which can compare binary files and give a result other then just 'differ' or 'same' the same as what diff does for text files. i though of using xxd with a regular diff, but that doesn't help, as if there is one byte missing, all lines are shifted which means they are different textually, when they are not really. -- One does not make peace with enemies. One makes peace with former enemies. My own blog is at http://www.zak.co.il/tddpirate/ My opinions, as expressed in this E-mail message, are mine alone. They do not represent the official policy of any organization with which I may be affiliated in any way. WARNING TO SPAMMERS: at http://www.zak.co.il/spamwarning.html = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: major packet loss at hot server
I don't have a machine that runs tcpdump. Plus it needs root access. I wonder if I create one of these free shells that are out there, will it help? tcpdump will work without root as well? Last time I talked with 012 and hot, hot managed to dissconnect me completely. The support from hot was nice, but told me to unplug the cable and replug again. And from then I don't have internet at all. The /etc/resolv.conf is correct. I waited for hot to call me back, but it didn't happen. Since now I am not at home, I can't check anything. When I will return back will have to shout at them to connect me to internet. All the time when I ran mtr google.com (outside) there was the packet loss. The loss is in the 2nd and 3rd hop. first is the router of hot ( I have another router at home but last time I didn't use it for the purpose of checking and proving hot it's not my router fault), then the switch with 80% packet loss and 3rd another switch/router of hot with 20% packet loss. 2nd hop is bgp. port 179 open. 3rd switch/router has port 49 open (tacacs). AND from my scannings, I pass only through 1 switch which has bgp open. I have access to a machine, but not as root. I can run there traceroute. And will definitely run from there mtr to machine A and traceroute from there to my machine. Are there any others programs that I can run as simple user? I can't understand them, if I pass through such a switch, where they close all the ports and allow only 179, 646, and all udp ports are closed what kind of internet is that. It castrates all the concept of internet. BTW, I want to ask a legal question. I will finally submit a complaint through tluna.co.il, but I don't know if it's legal to submit the ip numbers. A lot of people are using these switches, and people don't know about this problem. Thanks in advance for all the help. Hi Sara, If you have access to a machine somewhere that runs tcpdump, then use hping commands from your MPLS to that machine in order to see if the packet loss is occuring on the outgoing or on the return trip. That is, hping out 50 packets. Check to see how many of those get to the target and then check to see how many of those that got to the target got back to you. If there is a more packet loss on the return leg than on the outgoing leg I would suspect a routing problem. If there is more loss on the early hops of the outgoing leg then I suspect either congestion or physical transmission problems with the final repeater or router. How do I use different routing? Any idea of which ips should I put? You can't if you dont have a static IP. Well, I guess that you could check the routing to your first hop out that has a routed IP address by using various foreign traceroute web pages. If the routing problem is low enough and there is only one BGP router through which your packets can pass, then you might not be able to see the routing problem. I was able to see the routing problem that I had at Netvision last week because I was able to access my line from different Netvision border routers because Netvision is pretty big. It was really clear that one particular border router was getting bad information from an internal router. I know someone who is very nice at 012. She belongs to service dep, and she promissed to send someone who knows well unix/linux. I will see what I can do today. Otherwise, www.tluna.co.il is the answer. I put last year a complaint there and they started to call me (not vice versa). The problem was solved. There is another weird thing which I don't understand: the ip 213.57.43.199 shows IP address: 213.57.43.199 Reverse DNS:[Timeout] Reverse DNS authenticity: [Unknown] ASN:8584 ASN Name: UNSPECIFIED (Barak AS) I don't belong to barak. Someone at 012 told me they have agreements and they use each router on the way. Most suppliers have agreements for routing between them instead or or as well as routing through the IIX. In addition many suppliers purchase overseas bandwidth from BBL or Barak/Netvision. You can probably ask Hot if they use Barak/Netvision. - yba http://www.dnsstuff.com/tools/ipall.ch?domain=213.57.43.199 If someone can explain this, I will be glad. On 3/29/08, Jonathan Ben Avraham [EMAIL PROTECTED] wrote: Hi Sara, Sounds like you should consider switching suppliers. Regarding my problem with Netvision reported last week, I was able to show Netvision the difference in results (5% loss vs 70% loss) when using different routes through Netvisions AS, mainly depending on where the connection originated but this did not actually prove that there was a routing error AFAIK. For some reason I didn't think to test if the packet loss was symmetric (outgoing as well as incomming). That would probably have been as close as you could get to a proof
Re: major packet loss at hot server
Oh, I forgot, I have static ip. On 4/1/08, Jonathan Ben Avraham [EMAIL PROTECTED] wrote: On Sun, 30 Mar 2008, sara fink wrote: Date: Sun, 30 Mar 2008 09:48:41 +0300 From: sara fink [EMAIL PROTECTED] To: Jonathan Ben Avraham [EMAIL PROTECTED] Cc: linux-il@cs.huji.ac.il Subject: Re: major packet loss at hot server I can't check symmetry. I don't have adsl. In my case the problem starts at hot. the first 2 hops belong to hot and the 3rd hop is 012. The 80% packet lost is in the first hop. The 2nd hop another 20%. I have MPLS without dialer. Hi Sara, If you have access to a machine somewhere that runs tcpdump, then use hping commands from your MPLS to that machine in order to see if the packet loss is occuring on the outgoing or on the return trip. That is, hping out 50 packets. Check to see how many of those get to the target and then check to see how many of those that got to the target got back to you. If there is a more packet loss on the return leg than on the outgoing leg I would suspect a routing problem. If there is more loss on the early hops of the outgoing leg then I suspect either congestion or physical transmission problems with the final repeater or router. How do I use different routing? Any idea of which ips should I put? You can't if you dont have a static IP. Well, I guess that you could check the routing to your first hop out that has a routed IP address by using various foreign traceroute web pages. If the routing problem is low enough and there is only one BGP router through which your packets can pass, then you might not be able to see the routing problem. I was able to see the routing problem that I had at Netvision last week because I was able to access my line from different Netvision border routers because Netvision is pretty big. It was really clear that one particular border router was getting bad information from an internal router. I know someone who is very nice at 012. She belongs to service dep, and she promissed to send someone who knows well unix/linux. I will see what I can do today. Otherwise, www.tluna.co.il is the answer. I put last year a complaint there and they started to call me (not vice versa). The problem was solved. There is another weird thing which I don't understand: the ip 213.57.43.199 shows IP address: 213.57.43.199 Reverse DNS:[Timeout] Reverse DNS authenticity: [Unknown] ASN:8584 ASN Name: UNSPECIFIED (Barak AS) I don't belong to barak. Someone at 012 told me they have agreements and they use each router on the way. Most suppliers have agreements for routing between them instead or or as well as routing through the IIX. In addition many suppliers purchase overseas bandwidth from BBL or Barak/Netvision. You can probably ask Hot if they use Barak/Netvision. - yba http://www.dnsstuff.com/tools/ipall.ch?domain=213.57.43.199 If someone can explain this, I will be glad. On 3/29/08, Jonathan Ben Avraham [EMAIL PROTECTED] wrote: Hi Sara, Sounds like you should consider switching suppliers. Regarding my problem with Netvision reported last week, I was able to show Netvision the difference in results (5% loss vs 70% loss) when using different routes through Netvisions AS, mainly depending on where the connection originated but this did not actually prove that there was a routing error AFAIK. For some reason I didn't think to test if the packet loss was symmetric (outgoing as well as incomming). That would probably have been as close as you could get to a proof of routing error. Does your packet loss depend on where you enter 012 from (i.e. from Med1, from the IIX, from 012 ADSL)? Is the packet loss symmetric? That is do you lose the same percent of packets on packet going out as comming in? From my experience with Netvision, the level of service that you get at the ISP depends on who knows you, or who knows someone who knows you. Shavua Tov, - yba On Fri, 28 Mar 2008, sara fink wrote: Date: Fri, 28 Mar 2008 16:43:37 +0300 From: sara fink [EMAIL PROTECTED] To: Jonathan Ben Avraham [EMAIL PROTECTED] Cc: linux-il@cs.huji.ac.il Subject: Re: major packet loss at hot server I haven't solved the problem yet. From 012 someone superior (network dep) are supposed to call me and they will put hot on conference and this time I intend to request the net admin/integrator to take care of that AND ask them to go directly to the switch and disable the firewall.The ip from where there is 75-80% packet lost is a principal switchand another 1 or 2 ips where I get additional ~15-20% packet lost. They have a harsh firewall on the main switch. ALL UDP ports are blocked. TCP also a lot of ports closed. tcptraceroute (as opposed
Re: binary diff
On Thu, Apr 3, 2008 at 3:25 PM, Omer Zak [EMAIL PROTECTED] wrote: xxd's function is exactly to convert from binary into text. Try to convert your binary files using: xxd -p -c1 (works for me: xxd -v output is xxd V1.10 27oct98 by Juergen Weigert). I checked also how to do it using od. The following would work (my od --version is 5.97): od -v -w1 -ta --address-radix none or: od -v -w1 -tx1 --address-radix none --- Omer As i replied to someone before, it works for small files. for big files it takes ages to diff using jdiff (jojo's diff) takes very short time, but it gives information of the diff sizes and not about the actual data On Thu, 2008-04-03 at 11:37 +0300, Yaacov Fenster - System Engineering Troubleshooting and other stuff wrote: Erez - Perhaps you could first convert the binary files to text representation (od --address-radix none) and then preform the comparison in that dimension? Yaacov Erez D wrote: hi i am looking for something like 'diff' which can compare binary files and give a result other then just 'differ' or 'same' the same as what diff does for text files. i though of using xxd with a regular diff, but that doesn't help, as if there is one byte missing, all lines are shifted which means they are different textually, when they are not really. -- One does not make peace with enemies. One makes peace with former enemies. My own blog is at http://www.zak.co.il/tddpirate/ My opinions, as expressed in this E-mail message, are mine alone. They do not represent the official policy of any organization with which I may be affiliated in any way. WARNING TO SPAMMERS: at http://www.zak.co.il/spamwarning.html = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Replacing kernel 2.4 - 2.6
Thank you all for your answers. Although I was hoping for a different answer (where's the magic dust when you need it), you guys at least saved me a few days of what will most likely be futile work. - Aviram = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Replacing kernel 2.4 - 2.6
On Thursday 03 April 2008, Geoffrey S. Mendelson wrote: On Wed, Apr 02, 2008 at 08:13:19PM -0200, Oleg Goldshmidt wrote: On the device front, the 2.6 kernel has much more knowledge of the device internals (a necessity driven, in particular, for laptops). Hotplugging is the norm rather than an exception - the kernel no longer differentiates between a device discovered at boot or a device added later. The kernel now is even able to know of devices whose drivers have not been loaded yet. The infrastructure dealing with boring stuff like reference counting, power management, etc., has been unified. This does have implications for userspace - since the kernel operates differently userspace may need to do much less work now and do it differently than before. The other day I upgraded a system by using YUM from Fedora Core 5, to Fedora 8. I don't know whether it was a change in the kernel or a change in the way the Fedora team did something but my hard drives changed from /dev/hdx to /dev/sdx. They were both 2.6 kernels. The change was easily acommodated once I figured out that what had happened and why it booted fine with the FC5 kernel, but paniced with no root device found with the F8 one. I only had to change /etc/fstab and /etc/rc.d/rc.local, but I was lucky. :-) I ran into the same problems with two recent 2.6.x kernels - one of them from Linus with my own home-grown .config and the other the Mandriva distribution kernel. See this bug report for more information: https://qa.mandriva.com/show_bug.cgi?id=35432 I also had to tweak /boot/grub. Regards, Shlomi Fish - Shlomi Fish [EMAIL PROTECTED] Homepage:http://www.shlomifish.org/ I'm not an actor - I just play one on T.V. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: binary diff
On Thursday 03 April 2008, Ely Levy wrote: There are a few of them, take a look for example at xdelta and bsdiff It really depends on what files you are planning to use the diff on. The more you know about the files, the better tool you can choose. I've had good experience with xdelta. IIRC, it works very similar to the way rsync transfers files, with an advantage to xdelta, since it works locally. Also, xdelta is available both for Windows and Linux, if that matters to you. --yuval signature.asc Description: This is a digitally signed message part.
Re: Replacing kernel 2.4 - 2.6
On Fri, Apr 4, 2008 at 4:43 AM, Aviram Jenik [EMAIL PROTECTED] wrote: Thank you all for your answers. Although I was hoping for a different answer (where's the magic dust when you need it), you guys at least saved me a few days of what will most likely be futile work. So was there any alternate path you took to get over the problem? --Amos
Re: IPv6 support in old (2.4) kernels
On Wed, Apr 2, 2008 at 9:28 AM, Rami Rosen [EMAIL PROTECTED] wrote: Hi, Following is a fifth comment: (better later than never) Hi Rami, Thanks for the pointer. It occurred to me that the IPv6 Stateless Address Autoconfiguration (SLAAC) RFC you mention here (RFC2462, from 1971 ) No, it is from 1998 - it is IPv6, after all. It obsoletes RFC1971 ;-) is obsolete; it was replaced by RFC 4862, from 2007. Right, from Sep 2007. The list of requirements I referred to dates from June 2007. I said the customer was very important, this means that timescales are important, too, so don't be surprised :). A meta-requirement is to follow the evolution of the IPv6 standards, so this is very helpful indeed Thanks again, -- Oleg Goldshmidt | [EMAIL PROTECTED] = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Replacing kernel 2.4 - 2.6
Aviram Jenik wrote: Thank you all for your answers. Although I was hoping for a different answer (where's the magic dust when you need it), you guys at least saved me a few days of what will most likely be futile work. - Aviram This is no less crazy than the other suggestions, but it may work. Replace the entire hardware with a new one (optional), install a fairly recent version of Linux, and Xen 2 on it. Copy your entire current RedHat 7.2 distro to the Xen image, and replace the kernel with a 2.4 kernel of your compiling which has the para-virtualization patch from the Xen 2 source tree (that's the reason you cannot use Xen 3). Hell, if might actually work on the same hardware you currently have. Shachar = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Replacing kernel 2.4 - 2.6
On Thursday, 3 בApril 2008, Shlomi Fish wrote: On Thursday 03 April 2008, Geoffrey S. Mendelson wrote: ... I don't know whether it was a change in the kernel or a change in the way the Fedora team did something but my hard drives changed from /dev/hdx to /dev/sdx. They were both 2.6 kernels. I ran into the same problems with two recent 2.6.x kernels - one of them from Linus with my own home-grown .config and the other the Mandriva distribution kernel. See this bug report for more information: https://qa.mandriva.com/show_bug.cgi?id=35432 The name change is a superficial result of a deeper change. The newer kernels replaced the old IDE drivers with libata (despite the name it's not a library, it's in the kernel). This layer unifies IDE/SATA/SCSI into a common layer and hence they all get the same naming. Initialy the kernels were shipping both drivers so that libata handled SCSI and SATA and the old driver was used for IDE. Later kernels totally dropped the old IDE drivers and now libata is used for all disk types -- as a result you have a uniform naming to all /dev/sd? As usual, Fedora tries to follow the (b)leading edge (e.g: like Debian testing), which is why you got hit first. BTW: this change was listed in in the Release-Notes: http://docs.fedoraproject.org/release-notes/f7/en_US/sn-OverView.html, which obviously no one reads ;-) Cheers, -- Oron Peled Voice/Fax: +972-4-8228492 [EMAIL PROTECTED] http://www.actcom.co.il/~oron ICQ UIN: 16527398 Linux: like the air you breathe, ubiquitous and free To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
