Harry (lists | 1 Apr 10:12 2001

Permissions Problem with 1.5-pre5 (rpm) on Mandrake Linux??? (help)

I recently installed the netatlk rpm on Mandrake 7.2 for a friend, and while
everything appears to install just fine, I can see the server, I can see the
voumes, 2-way encrypted works, but I can't copy any files to the server.

Everytime I do, I get the dreaded "You do not have sufficient permissions'
error.

The shared directories have been chmod 777, just to cover all the bases, as
well as chmod g+s. Legit user is logged in, and used for connection.

This has happened with keeping the shared volumes as ext2 volumes, as well
as ReiserFS volumes (including for RAID volumes, as well as standard
volumes).

I'm at a loss - any ideas?

Harry

Heath A. Kehoe | 2 Apr 21:02 2001
Picon

Re: Permissions Problem with 1.5-pre5 (rpm) on Mandrake Linux???(help)


"Harry (lists)" wrote:
> 
> I recently installed the netatlk rpm on Mandrake 7.2 for a friend, and while
> everything appears to install just fine, I can see the server, I can see the
> voumes, 2-way encrypted works, but I can't copy any files to the server.
> 
> Everytime I do, I get the dreaded "You do not have sufficient permissions'
> error.
> 

There's a known problem with pre5 that will cause that.  This problem has
been fixed in the development sources, and so the next release should work
better.  In the meantime, I recommend running one of the 1.4 releases, such
as 1.4b2+asun2.1.3.

That bug may have caused some files in your .AppleDouble directories to have
bad permissions.  I have a perl script that will fix that (as well as other
problems):  http://www.avalon.net/~hakehoe/cleanappledouble.pl

-heath

Harry (lists | 2 Apr 21:45 2001

Re: Permissions Problem with 1.5-pre5 (rpm) on Mandrake Linux???(help)

> There's a known problem with pre5 that will cause that.  This problem has
> been fixed in the development sources, and so the next release should work
> better.  In the meantime, I recommend running one of the 1.4 releases, such
> as 1.4b2+asun2.1.3.

Any news on when the next iteration of netatalk will be out, then :-)

> That bug may have caused some files in your .AppleDouble directories to have
> bad permissions.  I have a perl script that will fix that (as well as other
> problems):  http://www.avalon.net/~hakehoe/cleanappledouble.pl

Thanks. I'll try that as well.

Harry

jeff | 3 Apr 10:01 2001
Picon

Re: filed a bug against afpd via sourceforge

Soren Spies wrote:
> 
> just filed this via sourceforge against netatalk-1.5 ...
> 
> In etc/afpd/file.c, around line 1289 in the function afp_createid, ad_close
> is called with a char* pathname instead of a struct adouble*.  It should
> instead pass (by reference) the struct adouble which was returned by
> reference from ad_open a few lines above.
> 
> I just fixed a bug in our read-only-AppleSingle-supporting afpd server
> running on AIX (based on the 1.4b2 release) in which the server would
> disconnect any time an operation on the client required afp_createid.
> We're unsure why the problem turned up all by itself, but our random
> string pointers consistantly caused ad_close to close file descriptor 0,
> which if you aren't running debug mode, is the network connection to the
> client.

Thank you. I've just integrated your patch into the CVS tree.

jeff

Gjermund Gusland Thorsen | 3 Apr 16:13 2001

Routing two LANs over internal WAN

Proabably not the right mailinglist…

There are two LANs

192.168.1.0/24
and
10.52.1.0/24

and a linket connecting the two LANs(MicroGate Synclink WAN adapter) 10.50.215.188/30

But I can’t get them too work… I’ve probably looked too much at it… pls help.

Route 1 and ISP connection:
/sbin/route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.1.48    0.0.0.0         255.255.255.255 UH    0      0        0 eth1
10.50.215.189   0.0.0.0         255.255.255.255 UH    0      0        0 mgsl0
LegalIP         0.0.0.0         255.255.255.255 UH    0      0        0 eth1
LegalIPNet      0.0.0.0         255.255.255.248 U     0      0        0 eth1
192.168.1.0     0.0.0.0         255.255.255.0   U     0      0        0 eth1
10.50.215.188   0.0.0.0         255.255.255.252 U     0      0        0 mgsl0
127.0.0.0       0.0.0.0         255.0.0.0       U     0      0        0 lo
0.0.0.0         LegalIP         0.0.0.0         UG    0      0        0 eth1

Router 2:

/sbin/route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
(Continue reading)

Jonathan Paisley | 3 Apr 17:57 2001
Picon
Picon

One bytelocking problem resolved?

Hi,

[sorry for long message; possible fix to a problem included!]

While using CodeWarrior Pro 5 with source files on a netatalk share, I
discovered a problem where CW silently ignores opening a file.

I think I've seen other people mention this issue before, so I suspect
it's not new, nor limited to CW.

After some playing, I found that it only happens (for me) with a file that
doesn't have a .AppleDouble entry (e.g. I've created the file from unix).

Further toying reveals that CW requests a byte lock on the file after
opening it (from 0 -> 0x7FFFFFFF). When the .AppleDouble entry exists, all
is well. However, if there is no .AppleDouble entry, the following code is
exercised:

etc/afpd/fork.c:596
    if (length == 0xFFFFFFFF)
      length = BYTELOCK_MAX;
    else if (length <= 0) {
      return AFPERR_PARAM;
    } else if ((length >= AD_FILELOCK_BASE) &&
	       (ad_hfileno(ofork->of_ad) == -1))
      return AFPERR_LOCK;

length here is 0x7FFFFFFF, BYTELOCK_MAX is a macro that evaluates to
0x7FFFFFFF. AD_FILELOCK_BASE is 0x7FFFFFFE (on i386 linux, at least).

(Continue reading)

Simon Vogl | 4 Apr 09:24 2001
Picon

Netatalk file time problems

Hi,
I am tracking down a file time reporting problem with netatalk
(netatalk-1.4b2+asun2.1.4 - the only
version I got to compile on my Sun 2.8 server).

We have a set of files created while daylight saving was off. Now it is
on and I have 1 hour time
difference for these files - files created now don't show the
difference, of course.

While reading the code for file.c in afpd ('round line 120), I saw the
following:
         case FILPBIT_CDATE :
            if (!isad || (ad_getdate(adp, AD_DATE_CREATE, &aint) < 0))
                aint = AD_DATE_FROM_UNIX(st->st_mtime);
						^^^^^
As I get correct creation dates, ad_getdate works on my system, but I
thing this should read
st->st_ctime?! (same in directory.c)

Does anybody have an idea how DST conversion works on macs? Do I have to
adapt the file mod times
on the server for files where it was not in effect???

Thanks
Simon

Bob Rogers | 4 Apr 15:51 2001
Picon

Netatalk file time problems

   From: Simon Vogl <vogl <at> soft.uni-linz.ac.at>
   Date: Wed, 04 Apr 2001 09:24:31 +0200

   Hi,
   I am tracking down a file time reporting problem with netatalk
   (netatalk-1.4b2+asun2.1.4 - the only
   version I got to compile on my Sun 2.8 server).

   We have a set of files created while daylight saving was off. Now it
   is on and I have 1 hour time difference for these files - files
   created now don't show the difference, of course.

Hmm, I notice this too (same netatalk version with a MacOS 9 client).
Is this dependent on the MacOS version, I wonder?  That would make it
tough to fix.

   While reading the code for file.c in afpd ('round line 120), 

The CVS code looks the same, by the way . . .

   I saw the following:

	    case FILPBIT_CDATE :
	       if (!isad || (ad_getdate(adp, AD_DATE_CREATE, &aint) < 0))
		   aint = AD_DATE_FROM_UNIX(st->st_mtime);
						   ^^^^^

   As I get correct creation dates, ad_getdate works on my system, but I
   thing this should read st->st_ctime?! (same in directory.c)

(Continue reading)

Kadari Mayson | 4 Apr 16:19 2001
Picon

Netatalk and filenames greater than 31 chars

I know that this question must have been asked and answered a million 
times, but is there yet a solution to allow filenames to be exported 
that are greater than 31 characters?  I don't want to pooch my netatalk 
install with 1.5 because I read that there are several problems with it, 
but I will try it if it offers a solution for the 31 characters question.

tia
/km

Matthew Keller | 4 Apr 16:40 2001

Re: Netatalk and filenames greater than 31 chars

	There is currently no solution - No pooching necessary ;) (Although
1.5 is very nice)

On 2001.04.04 10:19:35 -0400 Kadari Mayson wrote:
> I know that this question must have been asked and answered a million 
> times, but is there yet a solution to allow filenames to be exported 
> that are greater than 31 characters?  I don't want to pooch my netatalk 
> install with 1.5 because I read that there are several problems with it, 
> but I will try it if it offers a solution for the 31 characters question.
> 
> tia
> /km
> 
> 
> _______________________________________________
> Netatalk-devel mailing list
> Netatalk-devel <at> lists.sourceforge.net
> http://lists.sourceforge.net/lists/listinfo/netatalk-devel
> 
> 
--

-- 
Matthew Keller
Enterprise System Analyst
Computing & Technology Services
Information Services Division
State University of NY at Potsdam

http://mattwork.potsdam.edu/

(Continue reading)


Gmane