Re: 20031125: ldm connection problems

>From: Unidata Support <address@hidden>
>Organization: Unidata Program Center/UCAR
>Keywords: 200311251543.hAPFh3EH027063 LDM request failure for workstation ETA

Mark and Charles,

I asked Mark:

re: What is the output from a notifyme command run on the client pointing
at the server?

>notifyme -h 198.206.38.13 -f EXP -o 600
>
>No output.

Using the login information you gave me to omega, I ran notifyme
to the upstream host and got:

notifyme -vxl- -f EXP -o 50000 -h 198.206.38.13
Nov 25 22:03:01 notifyme[95050]: Starting Up: 198.206.38.13: 20031125080941.622 
TS_ENDT {{EXP,  ".*"}}
Nov 25 22:03:01 notifyme[95050]: Connected to upstream LDM-5
Nov 25 22:03:51 notifyme[95050]: Connected to upstream LDM-5
 ...

This seems to indicate that:

- you are able to connect to the LDM on 198.206.38.13
- you are allowed to request data using the EXP feed type
- there is no data in the upstream machine's LDM queue

At the same time, however, ldmping to the same host indicates that
there is no LDM running on the upstream server, or more specifically,
that there no LDM server on the upstream machine is accessible either
through port 388 or through some other port controlled by the
portmapper.

One possibility is that the LDM was not installed on 198.206.38.13
with setuid root privilege.  If this is the case, then it will not
use port 388, rather, it will use a port assigned by the port mapper.
Since the port mapper is considered to be a security hole, it is
possible that 198.206.38.13 is either not running the port mapper,
or access to the port mapper is not allowed through some sort of
firewall configuration.

The next question is for Charles Mcgill:

Can you verify if your LDM was installed with setuid root privilege?
This is done by doing a long list of ~ldm/bin/rpc.ldmd and
~ldm/bin/hupsyslog.  Here is example output from Mark's machine at
Lyndon State:

ls -alt ~ldm/bin/rpc.ldmd
-rwsr-xr-x  1 root  udata  149308 Aug 19 19:30 bin/rpc.ldmd
ls -alt ~ldm/bin/hupsyslog
-rwsr-xr-x  1 root  udata  5128 Aug 19 19:30 /software/ldm/bin/hupsyslog

If either/both of this routines does not have the setuid root bit
set, then you should have your system administrator ('root') do the
following:

<login>
cd ~ldm/ldm-6.0.14/src
make install_setuids

If the source is not available, 'root' can accomplish the same thing
with:

cd ~ldm/bin
chown root rpc.ldmd hupsyslog
chmod u+srwx,g+rx,o+rx rpc.ldmd hupsyslog

Some other things to check on the LDM installation on the upstream machine
(198.206.38.13) are:

1) was /etc/syslogd.conf setup correctly and a HUP sent to syslogd
2) was the LDM entry added to /etc/rpc
3) was the LDM entry added to /etc/services

Back to Mark's replies to my questions:

re: request for output from ldmd.log file on client (downstream)
>
>Nov 25 15:45:57 omega 198.206.38.13[11917]: ERROR: requester6.c:459; 
>ldm_clnt.c:258: Couldn't connect to LDM 6 on 198.206.38.13 
>Nov 25 15:46:27 omega 198.206.38.13[11917]: Desired product class: 
>20031125144340.597 TS_ENDT {{EXP,  ".*_wseta"}} 
>Nov 25 15:48:42 omega 198.206.38.13[11917]: ERROR: requester6.c:459; 
>ldm_clnt.c:258: Couldn't connect to LDM 6 on 198.206.38.13 

OK, this shows that Mark's LDM is not able to connect to an LDM-6
on 198.206.38.13.  See comments above for possibilities.

Tom