The dump does not help me much. I need to reporduce the problem in a controlled environment.

Stefan

Hello!

We have a number of local ELOG servers, all mirrored to a single "remote" ELOG server.  We have users that create updates at the local server, and some at the remote server, which can run the risk of record conflicts.  Right now, the local servers perform a "Mirror cron" every 5 minutes, but even that leaves the door open to potential conflicts.

I found an open-source JAVA-based app called DirSync Pro (https://www.dirsyncpro.org/) which is capable of performing real-time mirroring, and has conflict handling.  I "took it for a spin", and it does do what it claims.  However, because each ELOG server performs record "indexing", it doesn't recognize records that aren't part of the current list of records. Restarting the ELOG server obviously corrects that, but I was wondering if there is another way to get the server to recognize newer "remotely-generated" records without restarting the server.

As always, I'm appreciative for the outstanding working that has been done to make ELOG the great application that it is!

Hello Frank,

It seems, you are using the mirror function of elog. It should resolve conflicts by itself acording to the documented rules. (https://elog.psi.ch/elog/config.html)
As I don't use this function, I can't say how good it works.

If you don't want to use this function, I would suggest using the "Execute new | edit | delete" feature to trigger a script after each change of elog entries.
This script could itself run "rsync" or your sync solution to make the sync and
should afterwards call "killall -HUP elogd" on the remote to let elog re-read the config (and this sould also update the indices)
(see Server Configuration https://elog.psi.ch/elog/adminguide.html)

If you have a sync-solution, which itself permanently observes folders for changes and syncs it by itself,
It should have the option to run a command after sucessful sync or you need an other method to call "killall -HUP elogd" after sync.

Personally I would recommend the mirror function as it has a internal conflict resolution.
I hope this helps.

Best wishes,
Sebastian

PS: I don't know anything about your setup, but maybe there is a solution, where you don't need the local servers.
As I think, the mirror function is mainly for backup reasons of a main server to a secondary one or similar.

Hi Sebatian,

Thank you for taking the time to answer...very much appreciated!

Although I'm running Windows OS, I do understand your approach, and will work on an analogous solution.

Our setup is interesting -- we're running many temperature chambers, each one having a dedicated computer running a unique instance of ELOG, each of which is regularly updated to let users know what the progress is for the lengthy temperature testing.  All of these individual logbooks regularly synchronize to a common 'mirror' ELOG server, which shows all the logbooks in one location.  Users can view all the logbooks on one screen, by connecting to the mirror server.  Since this can be done remotely, it also makes it convenient to add "updates" remotely to the mirror server, which eventually synchronizes with each individual computer at the temperature chambers.  As you might imagine, if there is a user doing an update at the temperature chamber computer while another user enters an update remotely to the mirror server, there is a chance of having a record conflict.

I was trying to avoid conflicts by having real-time mirroring, where a change on either end is detected and immediately "synchronized", thereby reducing or eliminating conflicts.

In any case, if I do come up with a good solution for Windows, I'll be sure to share what I did.

Cheers,
Frank

Hi Frank,

I am not sure, if I understood your setup correctly. But in my eyes, you don't need the local elog servers. The only difference for the users at the chambers would be to directly use the 'mirror' remote elog url instead of the local elog url in their browsers. "which is regularly updated..." could mean, that you are using some kind of automatism to add entries to the local elogs, but you could also use these directly on the remote 

If you have concers about users editing the wrong chamber elog, there is a usermanagement to only allow certain users to certain elogs (on the remote).

As for the part "having a record conflict". That would be totally fine and handled by the mirror function.
See the part "If new entries exist locally and remotely having the same entry ID"... in the documentation of the function.

As you state, that you are working on windows. There is no "killall" command to send the signal, as far as i know.
Then a script could kill the elog and start it again. But this should have the disadvantage of loosing the login session of the users.

Best wishes,
Sebastian

Hi Sebastian,

You're absolutely correct that the users at the chambers could directly use the remote ELOG server (without having a local server), and I did originally think about this.  Unfortunately, there are times that our network "goes down" (for maintenance and other issues), and it was important to maintain the users' ability to perform regular 'local' updates at the temperature chambers, regardless of the network status.

In our situation, having a remote server also became particularly useful in the event that the 'local' computer at a specific temperature chamber "went down".  The user can continue to update the respective logbook at the remoter server, knowing that it would eventually synchronize with the respective local server(s).

Thanks for the great feedback!
Frank

Hi,

I am having trouble with uploading large (>0.5MB) files to elog. We click on upload and in a couple of seconds, the webbrowser complains that the server has dropped the connection.

Following the suggestions I found on these forums (https://midas.psi.ch/elogs/Forum/66753), I increased the timeout.tv_sec to 30 in three locations in elogd.c, but this did not help.

The problem is present in my old elog installation (from ~2 years ago), as well as the latest git snapshot from bitbucket that I cloned on June 10, 2017.

PS: Upload seems to work for non-secure configuration. It still takes a while to load, but it completes. However we prefer to use secure connections ( SSL = 1 ).
PS: Using nmap I looked at the latency to the relevant port, it can be as high as 0.5sec, but most often it is shorter.

Cheers,
e.

Hi,

Could someone at least suggest how I could debug this problem myself? If I know where to start, perhaps I can fix it myself and contribute to the software.

Best,

e.