rsync hang, more details [LONG]

Ed Santiago esm at
Tue Dec 18 14:09:28 EST 2001

  >The other thing you absolutely need to send is the output of 
  >netstat -ta

Alas, this is Solaris... and not a system on which I have root.
I don't think this shows anything useful, but here goes.

Hosts are fido (client) and cvsroot (server).  Output of netstat:

    cvsroot$ netstat -a -P tcp |grep fido
       Local Address        Remote Address    Swind Send-Q Rwind Recv-Q  State
    -------------------- -------------------- ----- ------ ----- ------ -------
    cvsroot.nfsd         fido.1012             8760      0 24820      0 ESTABLISHED        fido.1023                1      0     0      0 ESTABLISHED
 >> cvsroot.988          fido.1022             8760      0 24820      0 ESTABLISHED

    fido$ netstat -a -P tcp |grep cvsroot
    fido.1012            cvsroot.nfsd         24820      0  8760      0 ESTABLISHED
    fido.1023              0    921  8760      1 ESTABLISHED
 >> fido.1022            cvsroot.988          24820      0  8760      0 ESTABLISHED

The ">>" indicate the rsync session.  The other two seem to have
been there a while, and stick around after I've killed off the
hung rsyncs.

The 8760/24820 numbers are meaningless: a netstat immediately after
starting rsync (even in the "receiving file list" phase) shows those
same numbers.

[aside] Interesting (?) data point: I was paying extra-close
attention to the "tail -f" of the logfile[1], and noticed it hanging.
It was stuck for maybe a minute at 39/939, then all of a sudden
blurted out 21 more files and hung again for about two minutes,
21 more files, hang again... this time for good ( > 5 minutes).
No idea if this is meaningful or useful, passed along w/o comment.

Thanks again,

 [1] log file is on NFS disk, but "tail -f" is running on same
     host, so it wouldn't appear to be NFS-related delays.

More information about the rsync mailing list