Bad problems downloading Knoppix 4.0 DVD

[jacksonon]

I tried downloading KNOPPIX_V4.0DVD-2005-08-16-EN.iso from http://ftp.chg.ru/pub/Linux/knoppix-dvd/ server and I failed.

There were two problems: While I was downloading the file connection failed. Upon reconnect Apache Web server on chg erroneously gave this file size and then I tried to countinue (wget -c) downloading file from ftp://ftp.chg.ru. This time `wget` overdownloaded file upto 3 402 235 928 bytes (I killed wget as soon I had noticed that wget exceeded original file size) with its original size only 3 324 309 504 bytes.

I hoped that file downloaded OK and issued this command on my downloaded file:

dd if=KNOPPIX_V4.0DVD-2005-08-16-EN.iso bs=4096 count=811599 | md5

Unfortunately computed md5 sum of the part of the file does not match with the original file.

Will anyone, please, provide this file md5sum for every 100K or 1M block OR give me advice how not to redownload the whole file? At first attempt I had only around 600Mb left to download when connection failed.

[Harry Kuhman]

It's not clear why you didn't just follow the advice give in these forums, in the Get Knoppix link, and in the Downloading FAQ to use Bittorrent, the problems you state would have been avoided, since BitTorrent will restart and will check the file segment by segment for you. But it is entirely possiable that Bittorent may do it for you anyway, if you arrange for it to download to the same directory where you put this corrupt file. My suggestion would be to install BotTorrent and make the needed changes to your firewalls and be sure that the BitTorrent ports are forwarded in your router to the computer in question. Then get the torrent and start the download. After you see that the files are starting to load in the new directory, stop BitTorrent, swap in a copy of your bad ISO file, and restart BotTorrent. With luck it should examine the new ISO, find the bad segments, and get the missing parts. This is just theory, of course, it's not really the way that BitTorrent is intended to work.

I don't have the time or inclination to slice the ISO up and make ton of md5s. If the above fails, have you used par2 files and quickpar? I might be convinced to make a par set of the iso that would let you determine how much you need to repair. But before I get your hopes up, let me make it clear that if you really needed 600 megs of pars to repair your ISO that there is no way that I could upload that much. A few meg, sure, 600 meg,sorry, but you could download the whole thing faster on BitTorrent than I could upload 600 megs anyway. Plus I've been using my upload bandwidth to seed BitTorrent ever since the ISO was released, and that would seem the better use of the bandwidth.

And it will take me a fair amount of time to compute the pars for a single 3.09 meg file, particularly on the old equipment that I have the ISO on. You might still be better off starting a new BitTorrent download; I expect that BitTorrent is so fast at this point (the number of seeders well outnumber the downloaders) that you cound get a new copy of the ISO before I could computer the par2 files. (By the way, BitTorrent starts slow but pick up speed quickly after a few minutes. Don't go by the first time estimates you see, let it run 10-15 minutes before you decide how long a download is going to take. And be sure that you have your router ports forwarded properly and firewall ports open if you hope to get decent speed at all.

[Harry Kuhman]

I thought I had responded to you pretty fast after you posted, but you didn't post back. If you went to sleep or work, I hope that you started the BitTorrent download before you did, you would likely have the ISO by the next time you check into your computer.

I've computed a minimal par2 set, based on 321 blocks of 10,368,000 bytes each. I can send you a URL for the minimal par2 file. That would be enough for you to use to determine how many good blocks you have. If you only need a couple of replacement blocks for quickpar to fix your ISO I'll make arangments to get them to you. If you need something in the order of that 600 meg then as I said I can't do it and it just doesn't make sense to go that route (also, the more recovery blocks you create, the longer quickpar takes, making 600 megs of quickpar recovery blocks for a 3.09 files would again be another thing that would take more time than downloading the iso from scratch on BitTorrent. Let me know how the BitTorrent repair approach works out and if you want the par2 file to check your ISO and give you a good/bad block count.

Also, one more thing: The mirrors suffer from a file corruption problem that BitTorrent does not. When I used the mirrors before BitTorrent downloads were available, about 1/2 of my downloads from Knoppix mirrors were bad, and the symptom resembles what you have seen, files larger than they should have been (although I never restarted a download). I even have some understanding of what was going wrong (I've discussed it in other threads, I will not bother to go into that now). If you do either the BitTorrent approach or the pars approach and find out that you have a lot more of your file bad than the last 600 megs, likely most of all of the 10 meg blocks, then you suffered from the same coruption that I used to see and you can not repair it. I have tested the par2 file I made against the 4.0 DVD-EN ISO and all blocks check out fine against a good download.

[Harry Kuhman]

jacksonon, what's your status?

I have par2 files for you if you want them (even if you have already downloaded another copy of the ISO it might be interesting to see if you can repair the old one with the pars files).

[jacksonon]

.

[Harry Kuhman]

I'm a little hard pressed to understand why you don't want to use the par2 files that I sent you a url for. If you do get someone to compute the md5 then you still have to figure out how to download just the parts that are bad, and you have to glue it all back together again. But if you run quickpar with the par2 files, it will tell you how many blocks are bad (I used slightly larger blocks but still under 2 meg in size, the information would still give you a good idea of where you stand) and it will repair the ISO for you if I provided enough missing blocks in the par files. But it does seem clear that you want to ignore this. Good luck getting someone to run your script.

[jacksonon]

I'm a little hard pressed to understand why you don't want to use the par2 files that I sent you a url for.

I would like you not to be upset but the par files you have suggested weigh extra 50Mb of very expensive traffic and I simply don't have a desire to download that thing again.

Anyway thank you very much, but one russian guy over chg.ru was faster to reply and he provided me with necessary md5 sums.

[Harry Kuhman]

I would like you not to be upset but the par files you have suggested weigh extra 50Mb of very expensive traffic and I simply don't have a desire to download that thing again.

I'm not upset, was just confused by no response. The file that was available for download was 50 meg for two reasons, I wanted to try to include enough recovery blocks to help, and I actually made it for another user with a somewhat different problem (proper size file, bad md5 sum, and a corrupt disk partition). I can post either just the tiny par2 file itself, or a par2 file with only one recovery block. Or either would be small enough to e-mail if you send me an e-mail addres by the Private Mesage feature near the top of this page (don't post it in the forums, the spam harvesters will get it that way).

Look for a private message soon that gives you a URL for just the minimal par2 file, and if it determines that you need just one or two blocks to repair the ISO I can upload only as many as you need, just let me know how many blocks it needs to do the repair.