Re: dd / netcat / md5sums: hashes do not match

From: solana (solana_nexus_at_hotmail.com)
Date: 09/18/03

  • Next message: solana: "Re: dd / netcat / md5sums: hashes do not match" 
    Date: 18 Sep 2003 12:42:21 -0700

    Casper H.S. *** <Casper.***@Sun.COM> wrote in message news:<3f69beed$0$58700$e4fe514c@news.xs4all.nl>...
    > solana_nexus@hotmail.com (solana) writes:
    >
    > >B# nc -l -p 6789 | dd of=./hda1.img bs=512 conv=noerror,sync
    > >A# dd if=/dev/hda1 bs=512 conv=noerror,sync | nc <B's IP> 6789
    >
    > I wouldn't use "sync" because sync causes blocks to be
    > padded with NULLs; this is certain to corrupt
    > the file.
    >
    >
    > >Still gives different md5sum results.
    >
    > >btw. the 'records' output also differs.
    >
    > >On A:
    > >3084416+0 records in
    > >3084416+0 records out
    >
    > >On B (the receiver):
    > >2323232+1140752 records in
    > >3463984+0 records out
    >
    > Each of the incomplete reads (1140752) will be zero filled and that's
    > why there's more output.
    >
    > Since it's over the net you probably get two 512 byte blocks and then
    > some each packet which accounts for the 2:1 complete/incomplete ratio.
    >
    > Perhaps you'd like to specify "obs" so that just the output
    > is blocked and aggregated.
    >
    > Casper
    >
    > >however, A is a live system. Cant unmount hda1 since it happens to be
    > >the / part. cant boot into single user since I then wouldnt have
    > >network conn, i.e. not be able to transfer the file.
    >
    > Ah, in that case the checksums will never match but as long as the
    > fs is mostly idle (kill syslog and other sources of logs); you
    > should get a near perfect copy; nothing fsck can't fix.
    >
    > Casper

    turns out that /dev/tty1 constantly changes its time since i am logged
    on via that tty. turned off syslogd, not using bash (bash_history
    changes all the time), md5sums still dont match.. actually, ran md5sum
    several times in a row just on machine A: hashes change all the time..

    aside from that, should i use obs on the receiver side only? MTU on
    both interfaces is 1500. any way I can eliminate the incomplete reads
    altogether? aside from the md5 issue..

    thanks!

  • Next message: solana: "Re: dd / netcat / md5sums: hashes do not match"
    • Quantcast