elogd -m , posted by Dave on Thu Mar 10 00:16:33 2016
|
I'm running ELOG on a seperate Windows Servers.
When I try to synchronize the logbooks using "elogd -m", I get an "Error sending local entry: Error transmitting message"
If I click on the "Synchronize" link in the logbook, it can correctly send entry to remote ELOG.
Can anyone tell me why I get this error?
==============================
Here is the command & output:
c:\Program Files (x86)\ELOG>elogd -m
Retrieving entries from "http://166.22.64.201/Equipment Log"...
All entries identical
Retrieving entries from "http://166.22.64.201/System Log"...
Error sending local entry: Error transmitting message |
Execute new|edit doesn't seem to work, posted by Juergen Diefenbach on Fri Feb 26 17:35:54 2016
|
I am trying to use the "Execute new|edit|delete" feature to track changes to an elog using git.
When a new entry is created a shell script should be executed (certain attributes of the logbook entry should be passed as arguments to it). The script will then automatically stage all changes and create an appropriate git commit message. For a "new" entry in the logbook the commit message is a bit more verbose. When editing or deleting a message only a short commit message like "Edit message 25" should be created. Therefore I really need to distinguish between "new", "edit", and "delete".
However, I experienced unreproducible behavior when creating "new" entries. Sometimes the command specified by "Execute new=" is called, sometimes the one specified by "Execute edit=" is executed. Most of the time it doesn't work as expected. 
What I did so far to understand what goes wrong is I looked into elogd.c and found invoid submit_elog(LOGBOOK * lbs) that bedit is sometimes true and sometimes false, when creating a new entry, although it should be false then (right?).
This is why elogd then sometimes retrieves the wrong shell command from getcfg(lbs->name, "Execute edit"...) instead of getcfg(lbs->name, "Execute new"...) around line 23610 (not exact due to some eprintf()s inserted here and there to look at bedit and so on).
When deleting an entry, everything works fine: the shell command defined via "Execute delete=" is run, but this happens somewhere else in the code so it's not a surprise.
Do you have an idea how to fix this or can you point me in some direction to further track it down? Thank you! |
Re: Execute new|edit doesn't seem to work, posted by Andreas Luedeke on Sun Feb 28 18:24:58 2016
|
| Juergen Diefenbach wrote: | I am trying to use the "Execute new|edit|delete" feature to track changes to an elog using git.
[...]
However, I experienced unreproducible behavior when creating "new" entries. Sometimes the command specified by "Execute new=" is called, sometimes the one specified by "Execute edit=" is executed. Most of the time it doesn't work as expected.
[...]
Do you have an idea how to fix this or can you point me in some direction to further track it down? Thank you! |
I'm just guessing, but you could try to disable the "save drafts" feature and see if the behaviour changes:
Save drafts = 0
If that is the case then Stefan will know where to look 
Cheers
Andreas |
|
Re: Execute new|edit doesn't seem to work, posted by Juergen Diefenbach on Mon Mar 7 13:03:13 2016
|
| Andreas Luedeke wrote: | I'm just guessing, but you could try to disable the "save drafts" feature and see if the behaviour changes:
Save drafts = 0
If that is the case then Stefan will know where to look
Cheers
Andreas |
Alright, I tried "Save drafts = 0" in the config file and now it seems to work all the time as expected!
Thank you, Andreas!
In principle I could live with this as I don't think I need to have drafts enabled.
Cheers,
Jürgen |
|
Re: Execute new|edit doesn't seem to work, posted by Andreas Luedeke on Tue Mar 8 14:12:26 2016
|
| Juergen Diefenbach wrote: |
| Andreas Luedeke wrote: | I'm just guessing, but you could try to disable the "save drafts" feature and see if the behaviour changes:
Save drafts = 0
If that is the case then Stefan will know where to look
Cheers
Andreas |
Alright, I tried "Save drafts = 0" in the config file and now it seems to work all the time as expected!
Thank you, Andreas!
In principle I could live with this as I don't think I need to have drafts enabled.
Cheers,
Jürgen |
I would speculate that "Execute new" is not executed, when a draft is saved (as one would expect, since it's only a draft), but is neither executed when a draft is submitted (not as one would expect).
Good that you're fine for the moment.
If you could add a minimal configuration to reproduce the problem, then Stefan would have it easier to debug it. |
Attribute Field Size Limited to 255 Characters, posted by Mike Bodine on Thu Mar 3 00:12:45 2016
|
I currently have an Attribute defined in my .cfg file as "Format <Attribute> = 0, attribname, attribvalue, 120, 500". I've found if I exceed 255 characters elogd.exe fails. Is there a method that will allow me to exceed 255 characters? |
|
Re: Attribute Field Size Limited to 255 Characters, posted by Andreas Luedeke on Thu Mar 3 14:51:55 2016
|
No. At least there is no simple method.
You would need to edit the source code and recompile elogd. But the number 256 is used for about 300 times in the source code, you'll have a hard time finding the right one to increase.
Cheers, Andreas
| Mike Bodine wrote: |
|
I currently have an Attribute defined in my .cfg file as "Format <Attribute> = 0, attribname, attribvalue, 120, 500". I've found if I exceed 255 characters elogd.exe fails. Is there a method that will allow me to exceed 255 characters?
|
|
|
Attribute Field Size Limited to 255 Characters, posted by Mike Bodine on Thu Mar 3 00:02:26 2016
|
I currently have an Attribute defined in my .cfg file as "Format <Attribute> = 0, attribname, attribvalue, 120, 500". I've found if I exceed 255 characters elogd.exe fails. Is there a method that will allow me to exceed 255 characters? |
Links to images in notification message, posted by Devin Bougie on Fri Feb 26 17:20:52 2016 
|
We see two problems with the notificaiton email when attaching an image inline using drag-and-drop.
- The thumbnail does not appear in its proper location inline. There is a ? placeholder where the thumbnail should be, and the thumbnail then appears at the end of the message.
- Neither the ? placeholder nor the thumbnail that do appear are clickable.
Our v2.9.2 test installation exhibits the desired behavior.
- The thumbnail appears in its correct location inline.
- If you click on the thumbnail, you arrive at a URL displaying the full size image.
I am not able to drag-and-drop into the body of this message to test the behavior here. I am attaching a screenshot showing an example of the problem from our server.
Any suggestions would be greatly appreciated.
Many thanks,
Devin
|
|
Re: Links to images in notification message, posted by Devin Bougie on Fri Feb 26 18:11:14 2016
|
From the notification I received, it looks like "2" is a problem with this server. In the email notification, we would like images to be a link to a URL displaying the full-size image (as they are in v2.9.2).
Thanks,
Devin |
Possible bug in elogd execute_shell, posted by Nigel Warr on Fri Feb 26 08:47:22 2016
|
I was just playing around with gcc6's new feature for warning about misleading indentation (which can often hide real bugs) and I think it found one in elog-3.1.1-1 at src/elogd.c:22538. Here there is an if statement, which looks as though it should be inside a loop, but it isn't. The code is:
for (i = 0; i < MAX_ATTACHMENTS; i++)
generate_subdir_name(att_file[i], subdir, sizeof(subdir));
if (att_file[i][0] && strlen(shell_cmd) + strlen(lbs->data_dir) + strl$
< sizeof(shell_cmd) + 1) {
strcpy(p, "\"");
strcat(p, lbs->data_dir);
strlcat(str, subdir, sizeof(str));
strlcpy(str, att_file[i], sizeof(str));
str_escape(str, sizeof(str));
strcat(p, str);
strcat(p, "\" ");
p += strlen(p);
}
and the if statment is accessing the loop variable i but it is actually outside the loop. Presumably, there should be some more curly brackets here. gcc6 gave the warning:
src/elogd.c: In function ‘execute_shell’:
src/elogd.c:22538:10: warning: statement is indented as if it were guarded by... [-Wmisleading-indentation]
if (att_file[i][0] && strlen(shell_cmd) + strlen(lbs->data_dir) + strlen(subdir) + strlen(att_file[i])
^~
src/elogd.c:22536:7: note: ...this ‘for’ clause, but it is not
for (i = 0; i < MAX_ATTACHMENTS; i++)
^~~
|
|
Re: Possible bug in elogd execute_shell, posted by Stefan Ritt on Fri Feb 26 09:09:03 2016
|
Absolutely correct! Nice to see compilers getting better and better. I changed the code and committed it.
| Nigel Warr wrote: |
|
I was just playing around with gcc6's new feature for warning about misleading indentation (which can often hide real bugs) and I think it found one in elog-3.1.1-1 at src/elogd.c:22538. Here there is an if statement, which looks as though it should be inside a loop, but it isn't. The code is:
for (i = 0; i < MAX_ATTACHMENTS; i++)
generate_subdir_name(att_file[i], subdir, sizeof(subdir));
if (att_file[i][0] && strlen(shell_cmd) + strlen(lbs->data_dir) + strl$
< sizeof(shell_cmd) + 1) {
strcpy(p, "\"");
strcat(p, lbs->data_dir);
strlcat(str, subdir, sizeof(str));
strlcpy(str, att_file[i], sizeof(str));
str_escape(str, sizeof(str));
strcat(p, str);
strcat(p, "\" ");
p += strlen(p);
}
and the if statment is accessing the loop variable i but it is actually outside the loop. Presumably, there should be some more curly brackets here. gcc6 gave the warning:
src/elogd.c: In function ‘execute_shell’:
src/elogd.c:22538:10: warning: statement is indented as if it were guarded by... [-Wmisleading-indentation]
if (att_file[i][0] && strlen(shell_cmd) + strlen(lbs->data_dir) + strlen(subdir) + strlen(att_file[i])
^~
src/elogd.c:22536:7: note: ...this ‘for’ clause, but it is not
for (i = 0; i < MAX_ATTACHMENTS; i++)
^~~
|
|
|
Possible bug in elogd execute_shell, posted by Nigel Warr on Fri Feb 26 08:38:06 2016
|
I was just playing around with gcc6's new feature for warning about misleading indentation (which can often hide real bugs) and I think it found one in elog-3.1.1-1 at src/elogd.c:22538. Here there is an if statement, which looks as though it should be inside a loop, but it isn't. The code is:
for (i = 0; i < MAX_ATTACHMENTS; i++)
generate_subdir_name(att_file[i], subdir, sizeof(subdir));
if (att_file[i][0] && strlen(shell_cmd) + strlen(lbs->data_dir) + strl$
< sizeof(shell_cmd) + 1) {
strcpy(p, "\"");
strcat(p, lbs->data_dir);
strlcat(str, subdir, sizeof(str));
strlcpy(str, att_file[i], sizeof(str));
str_escape(str, sizeof(str));
strcat(p, str);
strcat(p, "\" ");
p += strlen(p);
}
and the if statment is accessing the loop variable i but it is actually outside the loop. Presumably, there should be some more curly brackets here. gcc6 gave the warning:
src/elogd.c: In function ‘execute_shell’:
src/elogd.c:22538:10: warning: statement is indented as if it were guarded by... [-Wmisleading-indentation]
if (att_file[i][0] && strlen(shell_cmd) + strlen(lbs->data_dir) + strlen(subdir) + strlen(att_file[i])
^~
src/elogd.c:22536:7: note: ...this ‘for’ clause, but it is not
for (i = 0; i < MAX_ATTACHMENTS; i++)
^~~
|
Kerberos auth not available on 3.0 ?, posted by Maikolk Kein on Mon Jan 26 10:30:40 2015
|
Hi all !!!
i have just updated and 2.9.2-1 elog instance, and i have problems with the kerberos configuration.
I was surprised to see on the logs, that the 3.0 daemon was refusing to start because the system
config has kerberos listed as an authentication method. I saw on the logs that the current daemon
didnt have kerberos support compiled in.
Ive seen that there is ldap support, but couldnt find anything about configuring this auth sytem, does
amnyone have any example i could check ?
Did i have to use a patched version ? or compile it on my own ?
Regards |
|
Re: Kerberos auth not available on 3.0 ?, posted by Andreas Luedeke on Mon Jan 26 15:04:38 2015
|
Hi Maikolk,
I have elogd 3.0 running with kerberos: I had no problems with that.
Maybe you forgot to change the Makefile before compiling?
Uncomment line 28 to:
# flag for Kerberos support, please turn on if you need Kerberos
USE_KRB5 = 1
and then "make clean" and "make install" (or "make update").
Regards, Andreas
| Maikolk Kein wrote: |
|
Hi all !!!
i have just updated and 2.9.2-1 elog instance, and i have problems with the kerberos configuration.
I was surprised to see on the logs, that the 3.0 daemon was refusing to start because the system
config has kerberos listed as an authentication method. I saw on the logs that the current daemon
didnt have kerberos support compiled in.
Ive seen that there is ldap support, but couldnt find anything about configuring this auth sytem, does
amnyone have any example i could check ?
Did i have to use a patched version ? or compile it on my own ?
Regards
|
|
|
Re: Kerberos auth not available on 3.0 ?, posted by Maikolk Kein on Mon Jan 26 15:06:49 2015
|
No no, i used the rpm form the website... I was thinking on having the bianries compiled, im downloading the server install dvd
to replicate the environment and have the same functionality.
Thanks for the answer ;-)
Regards
| Andreas Luedeke wrote: |
|
Hi Maikolk,
I have elogd 3.0 running with kerberos: I had no problems with that.
Maybe you forgot to change the Makefile before compiling?
Uncomment line 28 to:
# flag for Kerberos support, please turn on if you need Kerberos
USE_KRB5 = 1
and then "make clean" and "make install" (or "make update").
Regards, Andreas
| Maikolk Kein wrote: |
|
Hi all !!!
i have just updated and 2.9.2-1 elog instance, and i have problems with the kerberos configuration.
I was surprised to see on the logs, that the 3.0 daemon was refusing to start because the system
config has kerberos listed as an authentication method. I saw on the logs that the current daemon
didnt have kerberos support compiled in.
Ive seen that there is ldap support, but couldnt find anything about configuring this auth sytem, does
amnyone have any example i could check ?
Did i have to use a patched version ? or compile it on my own ?
Regards
|
|
|
|
Re: Kerberos auth not available on 3.0 ?, posted by Stefan Ritt on Tue Jan 27 17:16:46 2015
|
Kerberos support is not compiled into the binary RPM. You have to comile from sources to get that. The reason is that there are different Kerberos libraries around, and different people need different ones.
/Stefan
|
|
Re: Kerberos auth not available on 3.0 ?, posted by Jimmy Bria on Tue Feb 23 22:32:30 2016
|
Hi Stefan - Do you know is Kerberos is available (compiled) on the Windows installer?
-Jimmy
| Stefan Ritt wrote: |
|
Kerberos support is not compiled into the binary RPM. You have to comile from sources to get that. The reason is that there are different Kerberos libraries around, and different people need different ones.
/Stefan
|
|
|
Re: Kerberos auth not available on 3.0 ?, posted by Stefan Ritt on Wed Feb 24 09:48:09 2016
|
Hi Jimmy, no Kerberos is not compiled into the Windows installer. You are the first one asking for that under Windows. All other Kerberos users use Linux so far. You have to compile the Windows version yourelf. This is because there are several versions of Kerberos around and people need differnt libraries. Sorry for that.
/Stefan
| Jimmy Bria wrote: |
|
Hi Stefan - Do you know is Kerberos is available (compiled) on the Windows installer?
-Jimmy
| Stefan Ritt wrote: |
|
Kerberos support is not compiled into the binary RPM. You have to comile from sources to get that. The reason is that there are different Kerberos libraries around, and different people need different ones.
/Stefan
|
|
|