Re: Unexpected behavior from cygpath command

2019-11-13 Thread Alfred von Campe
On Nov 13, 2019, at 2:08, Frank Redeker wrote:

> I think on both systems the handling of 8.3 names is configured
> differently. You can check this with the Window command fsutil. (This
> command requires elevated permissions)
> 
> I get the following output on my system.
> 
> C:\WINDOWS\system32>fsutil 8dot3name query d:
> The volume state is: 0 (8dot3 name creation is enabled).
> The registry state is: 2 (Per volume setting - the default).

Thanks, I think this is very interesting, I did not know that such a setting 
existed.  It was indeed disabled for my E: drive.  However, after enabling it I 
still can’t get “cygpath -d" to work as expected.  This setting also doesn’t 
explain why cygpath returns the correct DOS path when I pass it in a Unix style 
path instead of a Windows style path.

Alfred
--
Problem reports:   http://cygwin.com/problems.html
FAQ:   http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple



Unexpected behavior from cygpath command

2019-11-12 Thread Alfred von Campe
I have two almost identical build servers, but cygpath is not behaving as 
expected on one of them.  Here is the output from the “good” build server:

  $ cygpath.exe —version | head -1
  cygpath (cygwin) 2.11.2
  
  $ cygpath -d 'E:\Program Files (x86)\IAR Systems'
  E:\PROGRA~1\IARSYS~1

Cygpath has correctly converted the given path to DOS (8.3) format as expected. 
Here is the output from the “bad” build server.

  $ cygpath.exe —version | head -1
  cygpath (cygwin) 2.11.1

  $ cygpath -d 'E:\Program Files (x86)\IAR Systems'
  E:\Program Files (x86)\IAR Systems

Why is cygpath returning the same path passed in?  Oh wait, it’s running a 
slightly older version (2.11.1 vs 2.11.2).  Perhaps there was a bug in the 
earlier version.  Let me update the Cygwin installation and try again:

  $ cygpath —version | head -1
  cygpath (cygwin) 3.1.0

  $ cygpath -d 'E:\Program Files (x86)\IAR Systems'
  E:\Program Files (x86)\IAR Systems

WTF?  Why is it still doing this?  Can there be a global config setting that 
affects cygpath’s behavior?  Hmm, let me try a different approach:

  $ cygpath -d "$(cygpath -u 'E:\Program Files (x86)\IAR Systems')"
  E:\PROGRA~1\IARSYS~1

Hey, cygpath can convert to DOS paths on this system after all, just not when 
it’s given a Windows path.  Can anyone explain this behavior?

Alfred



--
Problem reports:   http://cygwin.com/problems.html
FAQ:   http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple



Is the Perl module Net::LDAP available for Cygwin's Perl implementation?

2013-06-25 Thread Alfred von Campe
If so, is there a way to install it from the command line?  I need it on a 
server that I usually just ssh into and running setup.exe is a little 
cumbersome.

If not, what is the best/easiest way to get it?  I would like to run a Perl 
script on a Windows server under Cygwin that requires Net::LDAP functionality.

Thanks,
Alfred


--
Problem reports:   http://cygwin.com/problems.html
FAQ:   http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple



Re: Strange symlink behaviour

2010-03-10 Thread Alfred von Campe

On Mar 10, 2010, at 4:37, Thomas Wolff wrote:

Actually, if the file system does not support hard links, a tar  
file that contains them doesn't properly extract at all, it will  
rather report errors for those files.
(At least that's what it does on a network filesystem whose mount  
doesn't support hard links either.)

So it's a problem to use hard links in tar for general use.


That's good to know.  In my limited experiments on the NAS box we use  
(which speaks both CIFS and NFS), extracting a tar file containing  
hard links from Windows via the Cygwin tar, appears to "do the right  
thing".  It even appears that it's an actual hard link when viewed  
from NFS (both files have the same inode).


Alfred


--
Problem reports:   http://cygwin.com/problems.html
FAQ:   http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple



Re: Strange symlink behaviour

2010-03-09 Thread Alfred von Campe

On Mar 9, 2010, at 13:44, Larry Hall (Cygwin) wrote:


On 3/9/2010 10:11 AM, Alfred von Campe wrote:
The solution was to use hard links in the tar file (which doesn't  
use up
any more space), and when it was untar'ed on Windows, everything  
worked

^
On NTFS partitions.  It will on any kind of FAT partition though.


I meant that it didn't take more space inside the tar file.  I don't  
mind if it takes twice the disk space once extracted.


Alfred


--
Problem reports:   http://cygwin.com/problems.html
FAQ:   http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple



Re: Strange symlink behaviour

2010-03-09 Thread Alfred von Campe
Here is an update from my experiments (sorry for the delay).  As  
Corinna explained, symlinks created in Cygwin are only recognized by  
Cygwin, so what I wanted to do wasn't really possible.  However, we  
still wanted to be able to update the share from Windows (a GUI  
application was written for the operator to update the share when a  
new release is available), so I really wanted to figure out a way to  
make this work on Windows.  The solution was to use hard links in the  
tar file (which doesn't use up any more space), and when it was  
untar'ed on Windows, everything worked as expected.


Thanks for all the help,
Alfred


--
Problem reports:   http://cygwin.com/problems.html
FAQ:   http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple



Re: Strange symlink behaviour

2010-03-04 Thread Alfred von Campe

Real symlinks on a remote CIFS system are not recognized by Cygwin or
Windows either.  Rather, they are converted to real files by the
underlying CIFS server so that "dumb" Windows clients don't get
confused.


Ah, that makes sense and explains some behavior I was seeing.


Symlinks created with and for Cygwin are not real symlinks (with an
exception).  They are files with special properties so that they are
recognized as symlinks by Cygwin.


Does that also apply to a symlink extracted from a tar file created  
on Linux?



Bottom line is, if you create symlinks via Cygwin, they will only be
recognized by Cygwin clients.  If you want symlinks which work on all
systems (NFS->Linux, CIFS->Windows, CIFS->Cygwin), you have to create
the symlinks using a Linux client.


That's what I was afraid of.  My real issue is that this all works  
when I extract the tar ball into an empty directory, but it doesn't  
when the symlink already exists and a short cut is created instead.   
If the "rm //path/to/share/*" would simply delete the symlink I would  
not be having this problem.  Wait, a thought just occurred to me.   
What if I delete the symlink first instead of deleting "*".  I'll try  
this when I get to work today and report back with the results.


Vielen Dank,
Alfred


--
Problem reports:   http://cygwin.com/problems.html
FAQ:   http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple



Strange symlink behaviour

2010-03-03 Thread Alfred von Campe
Do symbolic links work over CIFS?  I'm having a really strange  
problem with symlinks.  Here is the story.


We have a small NAS box that is accessed via both NFS and CIFS.  The  
former is needed for the actual production access of the files on the  
share, but the latter is used to update the contents of the share.   
That is, we need to access the files from Linux via NFS, but want to  
update the files on Windows via CIFS.  The problem is that we need to  
create a symlink on the share, and to make it easier/possible to  
update the share from Windows we create a tar file that contains the  
three files we need:


  file_with_version_v1.0.3.abc
  file.abc ->  file_with_version_v1.0.3.abc
  other_file.def

If the share we want to populate with these files is empty,  
everything works fine.  However, when we try to install the new  
release on top of an existing release, we run these commands:


  rn -fv //path/to/share/*
  tar xvf //path/to/temp/release.tar -C //path/to/share

If the new version is 1.0.4, I expect the following files in the  
share after running those commands:


  file_with_version_v1.0.4.abc
  file.abc ->  file_with_version_v1.0.4.abc
  other_file.def

But instead, we get this (when looked at via NFS):

  file_with_version_v1.0.4.abc
  file.abc ->  file_with_version_v1.0.3.abc
  file.abc.lnk
  other_file.def

Note that the link is pointing to the old file and there is a  
shortcut added to the directory. Upon further inspection, when we run  
the "rm -f //path/to/share/*" command the directory appears empty  
when checking it on Windows or via CIFS.  However, when accessing it  
via NFS, I can still see the symlink.  So when the tar file is  
untar'ed, the symlink does not get overwritten and a shortcut is  
created instead.


We are still running Cywin V1.5 on the manufacturing floor, and  
updating to V1.7 is not feasible at the moment.  I think Cygwin uses  
some special tricks to get symlinks to work on Windows, but I don't  
quite understand how it works.  Any ideas what may be going on?


Thanks,
Alfred


--
Problem reports:   http://cygwin.com/problems.html
FAQ:   http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple



Re: Strange problem with running dmake via ssh

2009-04-02 Thread Alfred von Campe

On Apr 2, 2009, at 13:30, Dave Korn wrote:

  Is this possibly caused by having $SHELL set in the environment  
in one case

but not the other?


We have a winner!  The SHELL environment variable was set in both  
cases to /bin/bash.  Well, doing a "echo $SHELL" returned "/bin/bash"  
in both cases.  However, doing a "env | grep SHELL=" only returned  
"SHELL=/bin/bash" in the ssh window.  If I unset the SHELL  
environment variable in the ssh window, I can get dmake to work!  I  
don't quite understand the subtle difference of how these environment  
variables are set, but I am happy that I can now run our builds over  
ssh.


So thanks for the SHELL environment variable hint which helped me  
find a solution, Dave.  I owe you a here>.


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: Strange problem with running dmake via ssh

2009-04-02 Thread Alfred von Campe

On Apr 2, 2009, at 11:23, Corinna Vinschen wrote:


This chapter in the new Cygwin 1.7 User's Guide is supposed to
explain what's going on when switching the user context from a
service like sshd:


Thanks, Corinna, I will read this as soon as I get a chance (it's  
waiting at the printer for me already).  I'm still running Cygwin  
1.5, but I guess the same principles apply.


One thing I forgot to mention is that we run the Cygwin sshd service  
as a specific user so that we can access network shares without  
having to specify a password, and I am logged in as that user.  I can  
run other Windows commands without a problem.  It's just dmake that  
is throwing a fit.


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: Strange problem with running dmake via ssh

2009-04-02 Thread Alfred von Campe
I have a strange problem when trying to compile a project under  
Cygwin with dmake.  If I start a remote desktop connection to our  
Windows 2003 server, and start a Cygwin window (i.e., run  
Cygwin.bat), I am able to build our project just fine.  This also  
works in a native CMD window.  However, if I ssh into the same  
server (using the same user account), when I try to run dmake I get  
the following error:


  $ dmake flashimage
  dmake .\trioimage.elf LINK_FILE=trioflashimage.ld  
BOOTING=BOOT_FROM_FLASH

  dmake:  Error -- \bin\bash: No such file or directory
  dmake.exe:  Error code 255, while making 'flashimage'

The dmake executable was provided by a 3rd party.  What's so  
different about running a bash shell under ssh than running a bash  
shell in a CMD window that would cause dmake to behave this way?   
We'd like to run remote builds automatically, so getting this to  
work with ssh is kind of important.


I tried running strace to figure out what's going on, but that  
doesn't work since dmake.exe is a Windows executable.  I also tried  
running dmake like this:


  $ cmd /C "dmake flashimage"
  dmake .\trioimage.elf LINK_FILE=trioflashimage.ld  
BOOTING=BOOT_FROM_FLASH

  dmake:  Error -- \bin\bash: No such file or directory
  dmake:  Error code 255, while making 'flashimage'

As you can see, it fails in the exact same way.  Are there some  
Windows executables that just can't be run in a Cygwin ssh bash  
shell?  I could understand if a command didn't run at all under  
Cygwin/bash, but I just don't understand why it runs in a local  
Cygwin shell window but not in an ssh Cygwin shell window.


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Strange problem with running dmake via ssh

2009-04-01 Thread Alfred von Campe
I have a strange problem when trying to compile a project under  
Cygwin with dmake.  If I start a remote desktop connection to our  
Windows 2003 server, and start a Cygwin window (i.e., run  
Cygwin.bat), I am able to build our project just fine.  This also  
works in a native CMD window.  However, if I ssh into the same server  
(using the same user account), when I try to run dmake I get the  
following error:


  $ dmake flashimage
  dmake .\trioimage.elf LINK_FILE=trioflashimage.ld  
BOOTING=BOOT_FROM_FLASH

  dmake:  Error -- \bin\bash: No such file or directory
  dmake.exe:  Error code 255, while making 'flashimage'

The dmake executable was provided by a 3rd party.  What's so  
different about running a bash shell under ssh than running a bash  
shell in a CMD window that would cause dmake to behave this way?   
We'd like to run remote builds automatically, so getting this to work  
with ssh is kind of important.


Thanks,
Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: rsync under cygwin too slow

2009-03-23 Thread Alfred von Campe
rsync -rtuplg --stats --verbose --progress --modify-window=2 / 
cygdrive/c/localfolder /cygdrive/z/backups/

...
...
The local and remote folders were already synchronized, that is why  
0 files were transferred. File list generation did take only 3  
minutes but the whole process took more than 1.5 hours to finish  
and consumed a whopping 1.9 GB of downloaded bytes and 0.4 GB of  
uploaded bytes as per network counters. For what? Zero files, Zero  
bytes of file data.


I'm pretty sure this is because rsync has to copy all the files  
locally to compare them to see if they need to be copied.  Try  
running the rsync as a service on the system that hosts the Z: drive  
(or run rsync over ssh to the system that hosts the Z drive) and you  
will see large performance improvement.



As far as I can see, rsync under cygwin is also too slow to be useful.


This has nothing to do with Cygwin or rsync.  If you ran ROBOCOPY  
under Windows using the same setup you would see the same performance  
penalty.


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: How do I run sshd as a particular user?

2008-04-11 Thread Alfred von Campe

On Apr 11, 2008, at 11:48, I wrote:

Well, I spoke a little too soon.  I got this working on two  
systems, but can not get it to work on a third.  The ssh daemon  
appears to start (neither cygrunsrv -S nor starting it from the  
Services Panel gives an error), but it really does not.


I managed to solve this by rebooting the system and re-running ssh- 
host-config (and then changing permissions, etc.).  I now have all  
three build systems working as expected.


Thanks again for all the help,
Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: How do I run sshd as a particular user?

2008-04-11 Thread Alfred von Campe

On Apr 2, 2008, at 9:27, Corinna Vinschen wrote:

- Did you check /var/run/sshd.log?  If it's empty it's probably  
because

  the domain user has no write permission.

- Does the domain user have an entry in the local /etc/passwd?  sshd
  needs that when checking file ownership.  And it allows to  
specify the

  user to cygrunsrv without the "domain\win_username" syntax.

- Did you chown /etc/ssh* and /var/empty to the domain user when  
trying

  to start the service under that account?  That's a must have.


It's taken me a while to get to this, but changing the above  
mentioned permissions did the trick.  I can now log on and access  
network shares!  This will make the build engineers' lives a lot  
easier.  Thank you for a great product and for all the help on this  
list!


Well, I spoke a little too soon.  I got this working on two systems,  
but can not get it to work on a third.  The ssh daemon appears to  
start (neither cygrunsrv -S nor starting it from the Services Panel  
gives an error), but it really does not.  The Event Viewer  
Application log shows the following:


  The description for Event ID ( 0 ) in Source ( sshd ) cannot be  
found. The local
  computer may not have the necessary registry information or  
message DLL files to
  display messages from a remote computer. You may be able to use  
the /AUXSOURCE=
  flag to retrieve this description; see Help and Support for  
details. The following
  information is part of the event: sshd: PID 2920: service `sshd'  
failed: signal 11

  raised.

Any ideas what could be raising signal 11 (SIGSEGV) and how I would  
go about debugging this?  ssh access was working to this machine  
before I changed the file system permissions and the account  
information for the sshd service.


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: How do I run sshd as a particular user?

2008-04-02 Thread Alfred von Campe

On Apr 2, 2008, at 9:27, Corinna Vinschen wrote:






Did you try anything besides switching the user?  For instance:

- Did you check the event log?


Yes, did not find anything useful.

- Did you check /var/run/sshd.log?  If it's empty it's probably  
because

  the domain user has no write permission.


No, I did not.


- Does the domain user have an entry in the local /etc/passwd?  sshd
  needs that when checking file ownership.  And it allows to  
specify the

  user to cygrunsrv without the "domain\win_username" syntax.


Well, there is an entry for a user of the same name.  Is that enough  
or is there a way to specify the fact that it's a domain user?


- Did you chown /etc/ssh* and /var/empty to the domain user when  
trying

  to start the service under that account?  That's a must have.


No, I did not know about this either.  I will try these suggestions  
as soon as I get a chance, but that might not be until I return to  
the office next week.  A quick question, though.  Instead of making  
all these changes, is there a better way to (re-)install Cygwin sshd  
so that it is properly set up for a domain user?


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: How do I run sshd as a particular user?

2008-04-02 Thread Alfred von Campe

On Mar 28, 2008, at 11:28, Dave Korn wrote:

Alfred von Campe wrote on 28 March 2008 12:30:


I have tried various ways to get the sshd service started as a domain
user (instead of the default local user "sshd_server") and can not
get it to work.  What is the correct syntax to specify a domain user
with cygrunsrv?  This is what I have tried:

   cygrunsrv -I sshd -u "DOMAINNAME\USERNAME" -w PASSWORD -d "CYGWIN
sshd" -p /usr/sbin/sshd -a -D -e "CYGWIN=bin tty smbntsec" -y tcpip


  That's the windows domain user syntax sure enough.  There aren't  
any shell

metacharacters in the password by any chance are there?


Nope, just upper and lowercase letters, numbers, and a dash.  I also  
ensured that the user had all the user rights as described in the  
openssh.README file (well, all except for Increase Quota, which for  
some reason was not defined on this system, and must not really be  
required since the sshd_server account also did not have that right  
and it is able to start the service).  The result is the same, the  
service starts and immediately stops.  There is nothing obvious in  
the logs.  I am not really a Windows person, so I've been working  
with one of our IT guys on this, but he is out today and I will be  
out tomorrow and Friday, so this will have to wait until next week.


Again, the problem I am trying to solve is to be able to kick off  
builds remotely and automatically on this Windows server.  To do  
this, we need password-less login, and to that end, we have exchanged  
ssh keys and have this working.  However, by exchanging ssh keys the  
user is never fully authenticated on the domain, so there is no  
access to network drives.  Is there any other way to have  
passwordless ssh access yet still be fully authenticated on the  
domain?  I thought starting the service as a domain user would  
accomplish this, but alas, I have not been able to do that.  So if  
there is any other way to achieve our goal, I'd be happy to try it.


Thanks,
Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: How do I run sshd as a particular user?

2008-03-28 Thread Alfred von Campe
[I'm reposting this with a couple of corrections/clarifications and  
also to raise its visibility since I didn't get any responses last  
time :-)]


It's been a month since Larry Hall replied to my last post on this  
topic (thanks, BTW), and this issue has bubbled up to the top again.   
I have tried various ways to get the sshd service started as a domain  
user (instead of the default local user "sshd_server") and can not  
get it to work.  What is the correct syntax to specify a domain user  
with cygrunsrv?  This is what I have tried:


  cygrunsrv -I sshd -u "DOMAINNAME\USERNAME" -w PASSWORD -d "CYGWIN  
sshd" -p /usr/sbin/sshd -a -D -e "CYGWIN=bin tty smbntsec" -y tcpip


This successfully installs the service, and if I look at it in the  
Services control panel, it shows the correct username (DOMAIN 
\USERNAME), but if I try to start the service I always get the error  
"The Cygwin sshd service in Local Computer started and then  
stopped".  If I substitute sshd_server for the user and supply the  
correct password, the sshd service starts correctly.  But I want to  
start the service as a domain user so that I can access network  
shares and resolve some build issues with Visual Studio that are  
apparently caused by not being fully authenticated.


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: How do I run sshd as a particular user?

2008-03-26 Thread Alfred von Campe

On Feb 26, 2008, at 18:29, Larry Hall (Cygwin) wrote:





How did you make this change?  If you removed and reinstalled the  
service

with 'cygrunsrv' like the sshd configuration script does, then use the
'-W, --passwd ' flag.  Otherwise, specify the password in
Control Panel->Administrative Tools->Services->Cygwin sshd Logon  
properties

page.


It's not a month since Larry posted this (thanks, BTW), and this  
issue has bubbled up to the top again.  I have tried various ways to  
get the sshd service started as a domain user (instead of the local  
sshd_server user) and can not get it to work.  What is the correct  
syntax to specify a domain user with cygrunsrv?  This is what I have  
tried:


  cygrunsrv -I sshd -u "DOMAINNAME\USERNAME" -w PASSWORD -d "CYGWIN  
sshd" -p /usr/sbin/sshd -a -D -e "CYGWIN=bin tty smbntsec" -y tcpip


This successfully installs the service, and if I look at it in the  
Services panel it shows the correct username (DOMAIN\USERNAME), but  
if I try to start the service I always get the error "The Cygwin sshd  
service in Local Computer started and then stopped".  If I substitute  
sshd_server for the user and supply the correct password, the sshd  
service starts correctly.  But I want to start the service as a  
domain user so that I can access network shares and resolve some  
build issues with Visual Studio that are apparently caused by not  
being fully authenticated.


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: performance issue with sshd and cygrunsrv

2008-03-12 Thread Alfred von Campe
With regard to this performance problem, I've discovered that this  
is not just an sshd problem -- scripts/commands I run through cron  
also exhibit the same behavior.  (ie. find, curl, date, etc).  Is  
there something about the user context under which these services  
run that could be causing this?  Or does this point to cygrunsrv?


What does your PATH environment variable look like?  Make sure it  
doesn't contain any network paths.  As soon as I removed them, I got  
much better performance with ssh (I haven't used cron with Cygwin).


Alfred



--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



How do I run sshd as a particular user?

2008-02-26 Thread Alfred von Campe
I've read about the restrictions on accessing shares while logged  
into a Windows system with the Cygwin ssh daemon.  We are interested  
in this to do remote builds, and it would be nice to access network  
shares.  We only really need one user to be able to log in, so I  
thought I'd change the CYGWIN sshd service to run as that user.   
However, when I changed the service and tried to start it, I got the  
following error message: "The VYGWIN sshd servcice on Local COmputer  
started and then stopped."  Any ideas what's going on?


I tried to revert to having the service started by the .\sshd user,  
but I can't get that to work no either!  I think it's because I am  
using the wrong password.  How can I change or reset the password on  
that account?


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: Can not log in to recently enabled/configured sshd service

2008-02-08 Thread Alfred von Campe
A quick update on the my situation (I am the original poster).  After  
removing Cygwin and reinstalling it, re-running ssh-host-config, and  
making the home directory in /etc/passwd a local directory (i.e., not  
a network drive), I am now able to log into the Windows system from a  
Linux system.  This is great and will help us kick off automated builds.


However, in the remote ssh session, I am not able to access any  
network drives (home directories or otherwise), which I can easily do  
in a Cygwin window when accessing our build server via terminal  
services.  Is this a restriction of how sshd is implemented, or can I  
reconfigure something so I can access network drives from within ssh  
session?


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: Can not log in to recently enabled/configured sshd service

2008-02-07 Thread Alfred von Campe

On Feb 7, 2008, at 9:46, Dave Korn wrote:

  It all depends where you keep your home directory.  For a home on  
your local
HD, that's fine, but if you have home on a network share, you need  
'smbntsec',

since ntsec defaults to only cover local drives for speed.


Yes, our home directories are on network drives (NetApp shares).  I  
am seeing some very strange behavior at the moment with permissions,  
and will need to work with our IT department at some mutually  
convenient time.  This may not happen for a few days, and I'll report  
back with my findings once I've tried various options.


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: Can not log in to recently enabled/configured sshd service

2008-02-06 Thread Alfred von Campe
If you want this enabled indelibly just for the sshd service and  
you don't
want to go skulking around in the registry, you can re-run ssh-host- 
config
and specify 'smbntsec' as one of the settings when it asks you what  
you want
for the service.  This is a good thing to do because it allows you  
to be
specific about the settings each service gets.  Otherwise, if you  
want the

setting globally, just set it in the CYGWIN environment variable and
restart all Cygwin services.


I'm getting really strange behavior now from the Cygwin sshd  
service.  I'm going to completely redo the Cygwin installation by  
following these steps:


  1. Stop all services (I believe I only have sshd)
  2. Remove all services
  3. Remove Cygwin directory
  4. Remove all Cygwin Registry keys
  5. Reinstall Cygwin
  6. Rerun ssh-host-config

Is there anything I should pay close attention to when I go through  
this reinstallation/reconfiguration process?


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Re: Can not log in to recently enabled/configured sshd service

2008-02-06 Thread Alfred von Campe

On Feb 6, 2008, at 9:55, Dave Korn wrote:

  How do the perms on /home/av16209/.ssh/* look?  How do they look  
with

CYGWIN=smbntsec?


Interesting.  I had CYGWIN set to "binmode tty ntsec" according to  
some instructions I found by googling.  Here are the relevant results:


  bash-3.2$ CYGWIN=ntsec /bin/ls -ld ~/.ssh
  drwxr-xr-x 1 av16209 Domain Users 0 Feb  6 09:28 //solid/av16209/.ssh
  bash-3.2$ CYGWIN=smbntsec /bin/ls -ld ~/.ssh
  drwx--+ 1 av16209 Domain Users 0 Feb  6 09:28 //solid/ 
av16209/.ssh


So it appears that CYGWIN=smbntsec is what I want.  How can I  
reconfigure the sshd service to use that setting?


Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/



Can not log in to recently enabled/configured sshd service

2008-02-06 Thread Alfred von Campe
I have a pretty vanilla Cygwin installation on a Windows 2003 Server  
(SP1) system.  Yesterday, I tried to enable the sshd service by  
running the ssh-host-config script.  Everything appeared to work fine  
and I let the script create the two recommended user accounts.  I was  
also able to start the sshd service successfully.  However, I can not  
log in to the system.  With the minimal /etc/passwd file, when I try  
to log in via ssh using a user that is not in the /etc/passwd file,  
it will not take my password at all (that's to be expected).  I then  
added my account (as well as all the other accounts in our Windows  
domain) to the /etc/passwd file by running the following command:


  mkpasswd -l -d >> /etc/passwd

Now when I try to log in I get immediately disconnected as shown in  
this debug output output:


# ssh -v [EMAIL PROTECTED]
OpenSSH_3.9p1, OpenSSL 0.9.7a Feb 19 2003
debug1: Reading configuration data /home/av16209/.ssh/config
debug1: Reading configuration data /etc/ssh/ssh_config
debug1: Applying options for *
debug1: Connecting to gandalf [139.68.133.49] port 22.
debug1: Connection established.
debug1: identity file /home/av16209/.ssh/identity type 0
debug1: identity file /home/av16209/.ssh/id_rsa type 1
debug1: identity file /home/av16209/.ssh/id_dsa type 2
debug1: Remote protocol version 2.0, remote software version OpenSSH_4.7
debug1: match: OpenSSH_4.7 pat OpenSSH*
debug1: Enabling compatibility mode for protocol 2.0
debug1: Local version string SSH-2.0-OpenSSH_3.9p1
debug1: SSH2_MSG_KEXINIT sent
debug1: SSH2_MSG_KEXINIT received
debug1: kex: server->client aes128-cbc hmac-md5 none
debug1: kex: client->server aes128-cbc hmac-md5 none
debug1: SSH2_MSG_KEX_DH_GEX_REQUEST(1024<1024<8192) sent
debug1: expecting SSH2_MSG_KEX_DH_GEX_GROUP
debug1: SSH2_MSG_KEX_DH_GEX_INIT sent
debug1: expecting SSH2_MSG_KEX_DH_GEX_REPLY
debug1: Host 'gandalf' is known and matches the RSA host key.
debug1: Found key in /home/av16209/.ssh/known_hosts:101
debug1: ssh_rsa_verify: signature correct
debug1: SSH2_MSG_NEWKEYS sent
debug1: expecting SSH2_MSG_NEWKEYS
debug1: SSH2_MSG_NEWKEYS received
debug1: SSH2_MSG_SERVICE_REQUEST sent
debug1: SSH2_MSG_SERVICE_ACCEPT received
debug1: Authentications that can continue:  
publickey,password,keyboard-interactive

debug1: Next authentication method: publickey
debug1: Offering public key: /home/av16209/.ssh/id_rsa
Connection closed by 139.68.133.49

The home directory defined in the Cygwin /etc/passwd file is "// 
nasbox/av16209" which is the same home directory used on the Linux  
system that I initiated the ssh connection from (only there it's  
mounted as "/home/av16209").  I have got passwordless ssh access  
working from all my Linux systems, so that part is set up correctly.


Any ideas as to why this is failing on the Cygwin side?

Thanks,
Alfred


--
Unsubscribe info:  http://cygwin.com/ml/#unsubscribe-simple
Problem reports:   http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ:   http://cygwin.com/faq/