Hi Laurent,

Thanks, good point! I'm building a vanilla version (no extra options enabled as I'm using Webserver authentication).

3.15 RPM:

-rwxr-xr-x. 1 elog elog 1574768 Feb 26 17:29 elogd

bash-5.1$ ldd /usr/local/sbin/elogd 
    linux-vdso.so.1 (0x00007fff215f7000)
    libssl.so.3 => /lib64/libssl.so.3 (0x00007feff74c6000)
    libkrb5.so.3 => /lib64/libkrb5.so.3 (0x00007feff73eb000)
    libldap.so.2 => /lib64/libldap.so.2 (0x00007feff7384000)
    liblber.so.2 => /lib64/liblber.so.2 (0x00007feff7372000)
    libpam.so.0 => /lib64/libpam.so.0 (0x00007feff7360000)
    libstdc++.so.6 => /lib64/libstdc++.so.6 (0x00007feff7139000)
    libm.so.6 => /lib64/libm.so.6 (0x00007feff705c000)
    libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007feff7041000)
    libc.so.6 => /lib64/libc.so.6 (0x00007feff6e38000)
    libcrypto.so.3 => /lib64/libcrypto.so.3 (0x00007feff6a07000)
    libk5crypto.so.3 => /lib64/libk5crypto.so.3 (0x00007feff69ee000)
    libcom_err.so.2 => /lib64/libcom_err.so.2 (0x00007feff69e7000)
    libkrb5support.so.0 => /lib64/libkrb5support.so.0 (0x00007feff69d4000)
    libkeyutils.so.1 => /lib64/libkeyutils.so.1 (0x00007feff69cd000)
    libresolv.so.2 => /lib64/libresolv.so.2 (0x00007feff69b9000)
    libevent-2.1.so.7 => /lib64/libevent-2.1.so.7 (0x00007feff6960000)
    libsasl2.so.3 => /lib64/libsasl2.so.3 (0x00007feff6940000)
    libaudit.so.1 => /lib64/libaudit.so.1 (0x00007feff6912000)
    libeconf.so.0 => /lib64/libeconf.so.0 (0x00007feff6905000)
    /lib64/ld-linux-x86-64.so.2 (0x00007feff7572000)
    libz.so.1 => /lib64/libz.so.1 (0x00007feff68eb000)
    libselinux.so.1 => /lib64/libselinux.so.1 (0x00007feff68be000)
    libcrypt.so.2 => /lib64/libcrypt.so.2 (0x00007feff6884000)
    libgssapi_krb5.so.2 => /lib64/libgssapi_krb5.so.2 (0x00007feff682d000)
    libcap-ng.so.0 => /lib64/libcap-ng.so.0 (0x00007feff6822000)
    libpcre2-8.so.0 => /lib64/libpcre2-8.so.0 (0x00007feff6786000)
    
3.15 source vanilla build:

-rwxr-xr-x. 1 root root 1503896 Jul 15 09:31 elogd

bash-5.1$ ldd elogd
    linux-vdso.so.1 (0x00007fff60bbf000)
    libssl.so.3 => /lib64/libssl.so.3 (0x00007f8f83e44000)
    libstdc++.so.6 => /lib64/libstdc++.so.6 (0x00007f8f83c1b000)
    libm.so.6 => /lib64/libm.so.6 (0x00007f8f83b40000)
    libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f8f83b25000)
    libc.so.6 => /lib64/libc.so.6 (0x00007f8f8391c000)
    libcrypto.so.3 => /lib64/libcrypto.so.3 (0x00007f8f834e9000)
    /lib64/ld-linux-x86-64.so.2 (0x00007f8f83ef0000)
    libz.so.1 => /lib64/libz.so.1 (0x00007f8f834cd000)

Best,
Enrico

Just to add some points for others who may find this of use in future.

The hard coded number of entries options or Moptions can have is 100.  You can edit the code and recompile, but that would
not gain you many more before other problems concerning memory come in.

Options allow you to only select one from the list; Moptions allow multiple selections from the list.

As mentioned by Sebastian (previous poster) and  in my suggestion. I imagined that by Wafer 1060 (say), no new work would be 
being done on wafers 1001, ... 1010, so you could edit the config file and remove those (M)options.  It does not remove these 
wafer IDs from past records, simply that they can no longer be selected for new work to be recorded.    In that way the 
Moptions list remains short but allows for hundreds or thousands of WaferIDs,  ON THE ASSUMPTION that say only 50 (and certainly 
less than 100) are being worked on at any one time.

The numbers I chose here were random, it's more to highlight the principle rather than a prescription.

David.

> Just my 2 cents:
> 
> There is a hardcoded limit how many entries the Option list can have. Without looking into the source, I assume the limit also exists for MOptions.
> If you want more, you have to recompile elog with the changed limit.
> 
> We have used the normal Options attribute and a "Execute new"-script to alter the elog config for the Options list: to sort the list (5 last used entries on top, the rest alphabetical) and remove very old entries, which are not needed any more.
> Remark: if you change the elog.cfg, you have to tell elog to reload the cfg. e.g. using "killall -HUP elogd".
> 
> Alternatively, you can add javascript code via a html file and the attributes "Top text" or "Bottom text" to manipulate the input fields on the client side.
> 
> Both ways are a little bit hacky, but they work.
> Best wishes,
> Sebastian
> 
> > Thanks for you help. This is almost it. 
> > 
> > The problem is that the items are options and not freely closable numbers. In the end, with your solution, it will show you all of the previously put IDs which will be 1000s of entries for us. I think I will just put a convention that we have to write the numbers spread with a comma in a string 
> > field.
> > 
> > Thanks.
> > 
> > Best,
> > 
> > Nick 
> > 
> > 
> > > I have replied to this entry, because, for some reason I don't understand, if I reply to your latest entry, I am
> > > automatically logged out.  I tried this multiple times, and also on many other entries and had no issues other than
> > > entry 69787 - any reason for this, Stefan?
> > > 
> > > Anyway, what about MOptions? That appears to do what your example, and needs two lines in elog.cfg file:
> > > 
> > > Moptions WaferID = 1001, 1002, 1003, 1004, 1005
> > > Extendable Options = WaferID
> > > 
> > > I've done a couple of quick tests on a test logbook I keep for such experimentation, and it appears to do all 
> > > you have asked of it.  I added a new option 1006.  However, I found that one has to add that new one on its own, 
> > > let the entry become proper, and then edit the entry to add the other, existing, values.   If you tick entries and 
> > > also add a new one, then your new entry is all those listed on their own, that is you would get and new entry
> > > in the config file such as "1002 | 1004 | 1006", rather than just 1006
> > > 
> > > This is probably an result of an unexpected use of Moptions and extendable options, rather than a bug per se.
> > > 
> > > > Hey, 
> > > > 
> > > > thanks for your answer. I completely get your point. However, I think my question as not precise enough.
> > > > 
> > > > I would like to have a numeric input, but many at the same time. When I make a new post, I would like to have an attribute 'wafer_IDs' that specifies the list of wafers this process has been performed with. So for a single post I would like to have a list like this:
> > > > 
> > > > wafer_IDs = numeric value, numeric value, numeric value, extendable
> > > > 
> > > > Note: I am not referring here to the option. The numeric values are freely chooses numbers, the only this that varies from post to post is the number of numeric values put.
> > > > 
> > > > Let me make an example (If the attribute were a string this would be the equivalent):
> > > > 
> > > > 1st post: A process that was run with 3 wafers (ID: 1000, ID: 1001 and ID: 1002):
> > > > wafer IDs = 1000, 1001, 1002
> > > > 
> > > > 2nd post: A process that is run with 2 wafers (ID: 1000 and ID: 1002):
> > > > wafer IDs = 1000, 1002
> > > > 
> > > > The string solves the issue, but is not as nice as having directly a list of integers.
> > > > 
> > > > Thanks for your help!
> > > > 
> > > > Best,
> > > > 
> > > > Nick

Just my 2 cents:

There is a hardcoded limit how many entries the Option list can have. Without looking into the source, I assume the limit also exists for MOptions.
If you want more, you have to recompile elog with the changed limit.

We have used the normal Options attribute and a "Execute new"-script to alter the elog config for the Options list: to sort the list (5 last used entries on top, the rest alphabetical) and remove very old entries, which are not needed any more.
Remark: if you change the elog.cfg, you have to tell elog to reload the cfg. e.g. using "killall -HUP elogd".

Alternatively, you can add javascript code via a html file and the attributes "Top text" or "Bottom text" to manipulate the input fields on the client side.

Both ways are a little bit hacky, but they work.
Best wishes,
Sebastian

> Thanks for you help. This is almost it. 
> 
> The problem is that the items are options and not freely closable numbers. In the end, with your solution, it will show you all of the previously put IDs which will be 1000s of entries for us. I think I will just put a convention that we have to write the numbers spread with a comma in a string 
> field.
> 
> Thanks.
> 
> Best,
> 
> Nick 
> 
> 
> > I have replied to this entry, because, for some reason I don't understand, if I reply to your latest entry, I am
> > automatically logged out.  I tried this multiple times, and also on many other entries and had no issues other than
> > entry 69787 - any reason for this, Stefan?
> > 
> > Anyway, what about MOptions? That appears to do what your example, and needs two lines in elog.cfg file:
> > 
> > Moptions WaferID = 1001, 1002, 1003, 1004, 1005
> > Extendable Options = WaferID
> > 
> > I've done a couple of quick tests on a test logbook I keep for such experimentation, and it appears to do all 
> > you have asked of it.  I added a new option 1006.  However, I found that one has to add that new one on its own, 
> > let the entry become proper, and then edit the entry to add the other, existing, values.   If you tick entries and 
> > also add a new one, then your new entry is all those listed on their own, that is you would get and new entry
> > in the config file such as "1002 | 1004 | 1006", rather than just 1006
> > 
> > This is probably an result of an unexpected use of Moptions and extendable options, rather than a bug per se.
> > 
> > > Hey, 
> > > 
> > > thanks for your answer. I completely get your point. However, I think my question as not precise enough.
> > > 
> > > I would like to have a numeric input, but many at the same time. When I make a new post, I would like to have an attribute 'wafer_IDs' that specifies the list of wafers this process has been performed with. So for a single post I would like to have a list like this:
> > > 
> > > wafer_IDs = numeric value, numeric value, numeric value, extendable
> > > 
> > > Note: I am not referring here to the option. The numeric values are freely chooses numbers, the only this that varies from post to post is the number of numeric values put.
> > > 
> > > Let me make an example (If the attribute were a string this would be the equivalent):
> > > 
> > > 1st post: A process that was run with 3 wafers (ID: 1000, ID: 1001 and ID: 1002):
> > > wafer IDs = 1000, 1001, 1002
> > > 
> > > 2nd post: A process that is run with 2 wafers (ID: 1000 and ID: 1002):
> > > wafer IDs = 1000, 1002
> > > 
> > > The string solves the issue, but is not as nice as having directly a list of integers.
> > > 
> > > Thanks for your help!
> > > 
> > > Best,
> > > 
> > > Nick

Thanks for you help. This is almost it. 

The problem is that the items are options and not freely closable numbers. In the end, with your solution, it will show you all of the previously put IDs which will be 1000s of entries for us. I think I will just put a convention that we have to write the numbers spread with a comma in a string 
field.

Thanks.

Best,

Nick 


> I have replied to this entry, because, for some reason I don't understand, if I reply to your latest entry, I am
> automatically logged out.  I tried this multiple times, and also on many other entries and had no issues other than
> entry 69787 - any reason for this, Stefan?
> 
> Anyway, what about MOptions? That appears to do what your example, and needs two lines in elog.cfg file:
> 
> Moptions WaferID = 1001, 1002, 1003, 1004, 1005
> Extendable Options = WaferID
> 
> I've done a couple of quick tests on a test logbook I keep for such experimentation, and it appears to do all 
> you have asked of it.  I added a new option 1006.  However, I found that one has to add that new one on its own, 
> let the entry become proper, and then edit the entry to add the other, existing, values.   If you tick entries and 
> also add a new one, then your new entry is all those listed on their own, that is you would get and new entry
> in the config file such as "1002 | 1004 | 1006", rather than just 1006
> 
> This is probably an result of an unexpected use of Moptions and extendable options, rather than a bug per se.
> 
> > Hey, 
> > 
> > thanks for your answer. I completely get your point. However, I think my question as not precise enough.
> > 
> > I would like to have a numeric input, but many at the same time. When I make a new post, I would like to have an attribute 'wafer_IDs' that specifies the list of wafers this process has been performed with. So for a single post I would like to have a list like this:
> > 
> > wafer_IDs = numeric value, numeric value, numeric value, extendable
> > 
> > Note: I am not referring here to the option. The numeric values are freely chooses numbers, the only this that varies from post to post is the number of numeric values put.
> > 
> > Let me make an example (If the attribute were a string this would be the equivalent):
> > 
> > 1st post: A process that was run with 3 wafers (ID: 1000, ID: 1001 and ID: 1002):
> > wafer IDs = 1000, 1001, 1002
> > 
> > 2nd post: A process that is run with 2 wafers (ID: 1000 and ID: 1002):
> > wafer IDs = 1000, 1002
> > 
> > The string solves the issue, but is not as nice as having directly a list of integers.
> > 
> > Thanks for your help!
> > 
> > Best,
> > 
> > Nick

Hey Enrico,

Do you activate also all options with your build ? (pam/ldap/kb5/ssl)

Can you compare ldd command results on elogd binaries builded by yourself and the one from RPM ?

$ ldd /path/to/elogd

Also, size of both elogd files.

Regards

Sorry for posting again but something else came up.

Actually, building from source (elog-3.1.5-1.tar.gz) works just fine on OpenShift too.

The problem described below only happens when installing the packaged binary elog-3.1.5-20240226.el9.x86_64.rpm.

Best,
Enrico

Hello!

We're setting up ELOG on OpenShift. ELOG is installed on a Alma Linux 9 image. The container and the elog demo works fine executing the docker image locally.

When deployed on OpenShift, we get a weird response, that results in a 502 Bad Gateway. The broken response looks like:
 

# curl -v -H 'X-Forwarded-User: enrico.gamberini@cern.ch' https://psi-elog-container2-elisa-epdtdi.app.cern.ch/demo/

<html>redir</html>
HTTP/1.1 200 Document follows
HTTP/1.1 200 Document follows
Server: ELOG HTTP 3.1.5-23df00d
Content-Type: text/html;charset=ISO-8859-1
Set-Cookie: elmode=Summary; path=/demo; expires=Friday, 07-Dec-35 06:30:26 GMT;
Pragma: no-cache
Cache-control: private, max-age=0, no-cache, no-store
 

Notice the HTML tag before the HTTP header, as well the duplicate HTTP header.

I understand that it might be difficult to reproduce, but any input would be very welcome!

Thanks!

Best,
Enrico