I automatically install software on 3 machines in our lab, and at the end of the process, use eLOG to record relevant information for each machine (all running Windows XP) via the command line program elog.exe.  2 of the machines log their info correctly, but on third machine, I get the following error:

    Cannot retrieve host name: No error

The error is not very clear (i.e., the error is "no error"?), and furthermore, operationally, the error makes no sense to me, for the three machines are supposedly configured the same.  Additionally, the elog.exe used is stored on a network drive, as are the other batch files that all3 machines use, so it is not as a problem could creep in due to differing versions/configurations of files.

Just in case I am overlooking something, here's the command line invocation (I've X'd out sensitive information):

C:\>x:\@adat\tools\elog.exe -h max13 -l "Maxtee" -u xxxxxxxxxx xx1234   -a Tester=AULS_xxx_y    -a M
achine="Mercury 99"   -a "Module Type"=Controller   -a "Action Description=Configuration Change"   -
a Title="AULS upgraded to 1.2.7344 from [1.1.7343]."   -a "Card Count=998352"   -m X:\@ADAT\InterimP
F\MERCURY29.Blat_Msg1_PostBoot.TXT 1>X:\@ADAT\InterimPF\MERCURY29_eLog_CmdLineOutput1_(PostBoot).txt

Cannot retrieve host name: No error

Obvoiusly, there *must* be some configuration difference between the 2 working and 1 non-working machine, but I am at a loss to figure it out.

I am not sure which version of eLOG is being used, but the size of the elog.exe is 139,264. 

I suggest a new option be added to display the version information, or display the version info whenever eLOG is execute, or display the version info when "-?" option is used.

The error is produced by the C routine "gethostbyname". This routine tries to obtain the IP address of your host max13 and fails. The error "no error" is the error passed to elog from the underlying operating system, so please blame Windows for that. I guess you have a problem that on the machine you get the error you have some DNS problem, so this machine does not know max13. Try to "ping" max13 or do a "nslookup max13" to verify this. Maybe you have to add your domain after max13, or put this machine explicitly into \Windows\System32\drivers\etc\hosts.

I added some version information being displayed with the "-?" option. This feature will be contained in the next release.

Stefan:

Thanks for the analysis and information.  Why am I not surprised that the "No error" came from Windows!? 

I tried some of your suggestions with the following results:

1.  Ping ==>= works as expected on the failing machine; that is, can successfully ping "max13"  I also can ping successfully in the reverse direction; that is, from "max13" to my "client" machine.

2.  NSLookUp ==>does not work on any of the 3 machines.

3.  Added the computer name "max13" to HOSTS file; ==> observed no change in behavior (i.e., still fails)

I did notice that on the 2 machines that function as desired that there is no anti-virus software installed, while on the non-funcational machine there is A-V SW installed.  I am able to bring up eLOG via the browser on the non-functional machine, so am unsure if the A-V SW is a factor; that is, why does ELOG via browser work as expected, but not via command line. 

QUESTION:  does command line ELOG connect differently than ELOG through the browser?

I ask this question for perhaps the A-V SW *is* blocking some port or whatever, and I will need to ask our IT department how to remove such a restriction.

ELOG connects in the same way as the browser, but it is a different program. I have seen some firewalls which explicitly exclude browsers in firewall rules. So it might well be that yours blocks ELOG but not your browser. So give it a try and disable it.

Since both my machine and the Elog server PC are inside my company's "security zone", I am not running a firewall on my machine.

For what it's worth:  in trying to research the "No error" message, I did come across this information on MS's website; that is, their statement is:

The gethostbyname function retrieves host information corresponding to a host name from a host database.

Note  The gethostbyname function has been deprecated by the introduction of the getaddrinfo function. Developers creating Windows Sockets 2 applications are urged to use the getaddrinfo function instead of gethostbyname.

SOURCE:  http://msdn2.microsoft.com/en-us/library/ms738524.aspx

I installed elog 2.7.2 rev 2041 into /data/elog on Redhat ES 4 Update 4. This is my initial elogd.cfg file:

[global]
port = 80
Resource dir = /data/elog/elog
Logbook dir = /data/elog/elog/logbooks
Logfile = /data/elog/elog/logs/access_log
Logging level = 2
Resolve host names = 0
Usr = nobody
Grp = nobody

[demo]
Theme = default
Comment = General linux tips & tricks
Attributes = Author, Type, Category, Subject
Options Type = Routine, Software Installation, Problem Fixed, Configuration, Other
Options Category = General, Hardware, Software, Network, Other
Extendable Options = Category
Required Attributes = Author, Type
Page Title = ELOG - $subject
Reverse sort = 1
Quick filter = Date, Type

and start elogd - all is right with the world, everything is working properly. I shut down elog and try to add more logbooks using this template:
[A-B]
Subdir = cr_a-b
Comment = Change requests for Customers A-B
Menu commands = New, Reply, Download, Find, Last day
List Menu commands = New, Find
Display mode = summary

I create 14 of them, one for every 2 letters of the English alphabet and 0-9. The logbook name and Subdir change appropriately. When I start elog, it stays up for about 30 seconds, then crashes. No messages are recorded in syslog nor in the logfile. The only thing that looks suspicious are a couple lines in syslog:
Feb 15 17:11:05 iadopsutil04p elogd[23761]: elogd 2.7.2 built Feb 14 2008, 20:55:38 
Feb 15 17:11:05 iadopsutil04p elogd[23761]: revision 2041
Feb 15 17:11:05 iadopsutil04p elogd[23761]: FCKedit detected
Feb 15 17:11:05 iadopsutil04p elogd[23763]: Cannot restore original GID/UID.
Feb 15 17:11:05 iadopsutil04p elogd[23763]: Cannot remove pidfile "/var/run/elogd.pid" ; Permission denied
Feb 15 17:11:05 iadopsutil04p elogd[23761]: Server listening on port 80 ...

This is obviously the child process responsible for the highlighted lines and it's coming from the cleanup function. Why it's getting called is beyond me.
Ideas?

Feb 15 17:11:05 iadopsutil04p elogd[23761]: elogd 2.7.2 built Feb 14 2008, 20:55:38 
Feb 15 17:11:05 iadopsutil04p elogd[23761]: revision 2041
Feb 15 17:11:05 iadopsutil04p elogd[23761]: FCKedit detected
Feb 15 17:11:05 iadopsutil04p elogd[23763]: Cannot restore original GID/UID.
Feb 15 17:11:05 iadopsutil04p elogd[23763]: Cannot remove pidfile "/var/run/elogd.pid" ; Permission denied
Feb 15 17:11:05 iadopsutil04p elogd[23761]: Server listening on port 80 ...

The error message above come from the PID file /var/run/elogd.pid which cannot be removed. This file is only for information purpose, to tell any script which PID the elogd daemon has. If one elogd crashed in the past, the file might still be there although the program is not running, causing some error to show up, but which can normally be ignored.

The fact that elogd stays for about 30 seconds can only be contributed to the fact that you have very large logbooks, which are scanned during startup. You can check this by starting elogd interactively with the "-v" option. Then you will see the scanning process. The next thing is if you really get a crash, you can produce a core dump, so it can be analyzed to figure out where elogd has crashed. To my knowledge the only way you can do that is by having some logbooks with invalid data. Can you try your set-up with empty logbooks?

- Stefan

The mxml source code should be moved into the elog-*/src directory. This requires changing the Makefile to:

MXMLDIR = src/mxml

Reason for change: if you want to build several versions of elog and you unpack the distributions in the same top-level directory, the most recently unpackaged version will overwrite the mxml directory contents, possibly introducing bugs.

From your point of view that request is reasonable, but the reason of having mxml in a separate top level directory is that I use it in many different projects. If I would keep a copy of mxml in each project, I would have to update all of them manually if there is a bug. On the other hand the code of mxml changes very seldom. If you look at the history, it got ~5 changes last year, while elog itself had hundreds. And since I do not change the API, a bugfix in mxml should be to the benefit of all elog versions. It is like with other projects which depend on a number of packages. Like if you install Apache for example, you need a certain version of OpenSLL, a certain version of libcrypt, and so on.  So think of mxml like another package you have to combine with elog.

We use ELOG inside a local network, but we want to allow access also from outside the firewall.

The suggested solution from Elog Admin Guide worked fine until version ELOG V2.6.5-1844 (essentially without FCK Editor):

ssh -L 1234:your.server.name:8080 your.firewall.name
firefox http://localhost:1234/

Howerver, since the update to ELOG V2.7.1-2002 the FCK Editor uses absolute links to insert images.

An image uploaded from the local network will have an address like:

http://your.server.name:8080/logbookname/080207_101110/Picture.jpg

while the same image uploaded through the firewall tunnel will have a link like:

http://localhost:1234/logbookname/080207_101110/Picture.jpg

Obviously images uploaded from outside are not visible from inside and vice versa.

Is it possible to convince FCK Editor to use relative links?

PS
ELOG is great!!

ssh -L 1234:your.server.name:8080 your.firewall.name
firefox http://localhost:1234/

http://your.server.name:8080/logbookname/080207_101110/Picture.jpg

http://localhost:1234/logbookname/080207_101110/Picture.jpg

I fixed this in SVN revision #2038.

ssh -L 1234:your.server.name:8080 your.firewall.name
firefox http://localhost:1234/

http://your.server.name:8080/logbookname/080207_101110/Picture.jpg

http://localhost:1234/logbookname/080207_101110/Picture.jpg

Thank you for the quick support!

Hi

Im currently experimenting with Elog to see if we can use it to track maintenance work. But Ive come up against a problem, When I export the data to a csv file,

the date has a comma inserted between the day of the week and the rest of the date. This means that all the fields are then shifted to the right after the date.

This looks like a bug to me, the test records included in the demo does not exhibit this problem, until a new record is added.

cheers Richard

  What about exporting record separated by a semicolong ";" ?

I fixed that in SVN revision #2040, the comma is simply replaced by a space in the date string.

Hi Stefan

Thanks for the quick reply.

We seem to be running 2012 - Which may explain the problem - I will see with our system administrator why we are not running 2040. By the way I also had this problem when running the windows version.

Cheers.. Richard

Hi Stefan

Our IT administrator has just downloaded and installed the latest source code from the site and installed it. We still now have version 2012?

Where do we get version 2041 ?

Cheers,  Richard

The bleeding edge version you get always from our SVN server, but you have to recompile it yourself. For your convenience, I just made you a version 2.7.2-3 containing the latest source code.

Thanks Stefan

Everything works now.

Hello,

one question regard Options: For e. g. I am using the following command: Options Categories = NT, 2000, XP, Linux
Now I would like to delete the entry NT. But that would cause, that NT is no longer available in the search function. So I wonder if there is a way to deactivate an attribute?

Thanks!

Uwe

A deactivate option is not implemented. The only possibility you have is to enter the search string drictly in the url, like

http://<host>/<logbook>/?Category=NT

Thanks for the quick answer!

best regards,

Uwe

How can I make a cross reference from one entry to another entry in the same ELOG logbook?

Have a look at http://midas.psi.ch/elog/userguide.html, under the section "Adding stuff to a logbook". Basically you can write elog:<id>. At the moment, this works however only in the "plain" or "ELCode" encoding. For HTML encoding, I will add this functionality soon.

Thank you! The HTML functionality will be v. useful

I implemented this in SVN revision #2036.

Hello,

How do I proceed to have spell check in fck editor under linux / firefox ?

Kind regards

Stéphane

You have to talk to the FCKeditor guys about that. On their website they mention IEspell, but this works only for MS Internet Explorer. On their demo site, I see the spell check icon, but it's greyed out. In this forum, spell checking works for me using Firefox 2.0 under Windows after I installed the Firefox extension "United States English Dictionary" from here. Maybe this works for you also under Linux. There is however one problem: I get wrong words underlined in red, but I cannot get suggestions by clicking at the wrong words. For FCKeditro version 2.6+ however it was announced that there will be a new spell checker, so maybe you have to wait for their update.