ID |
Date |
Icon |
Author |
Author Email |
Category |
OS |
ELOG Version |
Subject |
67166
|
Wed Jan 25 14:05:46 2012 |
| Christian Herzog | herzog@phys.ethz.ch | Comment | All | 2.9.0 | Re: problems with https in Chrome and IE |
Andreas Luedeke wrote: |
Christian Herzog wrote: |
[...] we're evaluating elog right now at the Physics Department of ETH Zurich and I'm trying to come up with a good config. One of the first steps of course was to enable SSL/https. With http, all tested browsers work fine, but with https at least Google Chrome 16 and IE 9 do not get past the "unknown certificate" warning and I see "TCP connection broken" errors in the log file. Firefox however works fine. Same behavior on Linux, Mac and Windows (given the browser in question is available). elog server is running on Lucid.[...]
|
⇄
Detect language » English
If you want to use https you should know what a certificate is.
Certificates are used to encript the data, but at the same time they are used to identify the host.
ELOG is delivered with a self generated certificate.
This can be used to encript the data, but no certification authority knows this certificate, so nobody can guaratee that you are connected to the right host.
Most browsers will warn you, that nobody did and if you don't care you need to change the security settings of you browser to accept the connection anyway.
The proper way out of this is to buy a certificate from a certification authority. Or to switch off https. (See https://midas.psi.ch/elog/config.html#global SSL option)
|
we know about certificates, thank you 
The point is that it stops AFTER the point at which I tell the browser to accept the self-signed certificates. I now even got a CACert and the problem remains: FF works, Chrome and IE don't: https://phd-bkp-gw2.ethz.ch:8080/admin/
log says: TCP connection broken
thanks,
-Christian |
67167
|
Wed Jan 25 14:48:36 2012 |
| Andreas Luedeke | andreas.luedeke@psi.ch | Comment | All | 2.9.0 | Re: problems with https in Chrome and IE |
Christian Herzog wrote: |
Andreas Luedeke wrote:
|
Christian Herzog wrote:
|
[...] we're evaluating elog right now at the Physics Department of ETH Zurich and I'm trying to come up with a good config. One of the first steps of course was to enable SSL/https. With http, all tested browsers work fine, but with https at least Google Chrome 16 and IE 9 do not get past the "unknown certificate" warning and I see "TCP connection broken" errors in the log file. Firefox however works fine. Same behavior on Linux, Mac and Windows (given the browser in question is available). elog server is running on Lucid.[...]
|
  ⇄
Detect language » English
[...] The proper way out of this is to buy a certificate from a certification authority. Or to switch off https. (See https://midas.psi.ch/elog/config.html#global SSL option)
|
we know about certificates, thank you 
The point is that it stops AFTER the point at which I tell the browser to accept the self-signed certificates. I now even got a CACert and the problem remains: FF works, Chrome and IE don't: https://phd-bkp-gw2.ethz.ch:8080/admin/
log says: TCP connection broken [...]
|
 ⇄
Detect language » English
Sorry that I was mis-interpreting your question 
Unfortunately I don't know what's wrong with your set-up. I can confirm that I cannot access your logbook with "konquerer", but can access it with "firefox". The "konquerer" (on Scientific Linux 5.7) just gets timed out.
But I can access other SSL/https ELOGs with the konquerer. The problem only occurs with your logbook!
Therefore I would think it is a particular problem of your installation. I have three ideas how to isolate the problem:
- first, I would try to change to the standard port 443. Just in case it is related to some firewall, etc. problem.
- second, I would try another operating system than Ubuntu Lucid. It should work of course with Ubuntu, but if it still doesn't work with the other operating system then many things are already ruled out.
- third, I would try to set-up an apache webserver in front of ELOG. We have it here just for safety reasons. ELOG runs then on some special port and apache connects to it with a reverse proxy.
The latter is a little bit of work (about a day) if you never set-up apache before. Therefore I would try the other two, first.
Good luck!
|
67168
|
Wed Jan 25 15:08:53 2012 |
| Christian Herzog | herzog@phys.ethz.ch | Comment | All | 2.9.0 | Re: problems with https in Chrome and IE |
Andreas Luedeke wrote: |
Christian Herzog wrote: |
Andreas Luedeke wrote:
|
Christian Herzog wrote:
|
[...] we're evaluating elog right now at the Physics Department of ETH Zurich and I'm trying to come up with a good config. One of the first steps of course was to enable SSL/https. With http, all tested browsers work fine, but with https at least Google Chrome 16 and IE 9 do not get past the "unknown certificate" warning and I see "TCP connection broken" errors in the log file. Firefox however works fine. Same behavior on Linux, Mac and Windows (given the browser in question is available). elog server is running on Lucid.[...]
|
  ⇄
Detect language » English
[...] The proper way out of this is to buy a certificate from a certification authority. Or to switch off https. (See https://midas.psi.ch/elog/config.html#global SSL option)
|
we know about certificates, thank you 
The point is that it stops AFTER the point at which I tell the browser to accept the self-signed certificates. I now even got a CACert and the problem remains: FF works, Chrome and IE don't: https://phd-bkp-gw2.ethz.ch:8080/admin/
log says: TCP connection broken [...]
|
 ⇄
Detect language » English
Sorry that I was mis-interpreting your question 
Unfortunately I don't know what's wrong with your set-up. I can confirm that I cannot access your logbook with "konquerer", but can access it with "firefox". The "konquerer" (on Scientific Linux 5.7) just gets timed out.
But I can access other SSL/https ELOGs with the konquerer. The problem only occurs with your logbook!
Therefore I would think it is a particular problem of your installation. I have three ideas how to isolate the problem:
- first, I would try to change to the standard port 443. Just in case it is related to some firewall, etc. problem.
- second, I would try another operating system than Ubuntu Lucid. It should work of course with Ubuntu, but if it still doesn't work with the other operating system then many things are already ruled out.
- third, I would try to set-up an apache webserver in front of ELOG. We have it here just for safety reasons. ELOG runs then on some special port and apache connects to it with a reverse proxy.
The latter is a little bit of work (about a day) if you never set-up apache before. Therefore I would try the other two, first.
Good luck!
|
thanks for the fast resonse.
1) port 433 done. No change
2) compiled elog 2.9.0 on Squeeze and only reused the config file. No change: https://daduke.org:8443/
3) we can do that (and we will) no problem, but I'd like to get it working w/o apache nonetheless
speaking of reverse proxy: we'd like to hook elog to our LDAP server. As there's no LDAP binding built in, is there any way to use apache LDAP auth and then bind to that one?
thanks,
-Christian |
67232
|
Mon Apr 9 19:58:51 2012 |
| Achim Dreyer | ml10352@adreyer.com | Bug report | All | 2.9.1 | wrong version number in spec file |
localhost$ svn diff elog.spec
Index: elog.spec
===================================================================
--- elog.spec (revision 2446)
+++ elog.spec (working copy)
@@ -2,7 +2,7 @@
Name: elog
Summary: elog is a standalone electronic web logbook
-Version: 2.8.1
+Version: 2.9.1
Release: 1
License: GPL
Group: Applications/Networking
|
67285
|
Wed May 23 05:29:35 2012 |
| Tim Thiel | tt2005@thieleng.com | Question | All | 2.9.0 | How does SVN info get placed in source? |
In the elogd.c file nearly at the very top is the following line:
char svn_revision[] = "$Id: elogd.c 2411 2011-04-01 14:39:35Z ritt $";
How does this information get into the source file? Is it a script used when the svn changes are committed, or perhaps when they are checked out? Any details or pointers to info on the web would be greatly appreciated.
Thanks.
|
67286
|
Thu May 31 11:40:15 2012 |
| Stefan Ritt | stefan.ritt@psi.ch | Question | All | 2.9.0 | Re: How does SVN info get placed in source? |
Tim Thiel wrote: |
In the elogd.c file nearly at the very top is the following line:
char svn_revision[] = "$Id: elogd.c 2411 2011-04-01 14:39:35Z ritt $";
How does this information get into the source file? Is it a script used when the svn changes are committed, or perhaps when they are checked out? Any details or pointers to info on the web would be greatly appreciated.
|
This gets handled automatically by SVN. All you have to do is to put an "$Id$" in your text somewhere and enable the keyword ID. See for example
http://svnbook.red-bean.com/en/1.4/svn.advanced.props.special.keywords.html |
67296
|
Wed Jul 4 13:58:23 2012 |
| Richard Stamper | richard.stamper@stfc.ac.uk | Question | All | 2.9.2 | Number of conditional attributes |
Is there a limit on the number of conditions that can be simultaneously active in a log? When I activate more than 10 conditions I start to see side effects with other conditions being deactivated. I think this is due to the hard-coded array size of 10 for clist in the match_param function in elogd.c. If so, could this limit be increased?
Are there other limits on the number of conditions, or the length of condition names? If I've understood the code right, the _condition string holds a comma-separated list of the active conditions, so the 256 byte length of this will also put some limit on the number of conditions that can be active. |
67301
|
Fri Jul 13 10:45:49 2012 |
| Stefan Ritt | stefan.ritt@psi.ch | Question | All | 2.9.2 | Re: Number of conditional attributes |
Richard Stamper wrote: |
Is there a limit on the number of conditions that can be simultaneously active in a log? When I activate more than 10 conditions I start to see side effects with other conditions being deactivated. I think this is due to the hard-coded array size of 10 for clist in the match_param function in elogd.c. If so, could this limit be increased?
Are there other limits on the number of conditions, or the length of condition names? If I've understood the code right, the _condition string holds a comma-separated list of the active conditions, so the 256 byte length of this will also put some limit on the number of conditions that can be active.
|
Ups. I never imagined that someone would use more than 10 conditions. I'm not 100% sure if the problem is the clist array, but can you try to set it to 20 and see if it gets better? You can also send me your config file and I can try it myself. |