You are not logged in.
Pages: 1
Hi, how are you?
I know that this topic was already posted here, but none of the solution proposed on those posts worked for me.
I have GLPI connected to OCS, both on the same server (Opensuse11) I'm currently managing over 650 computers plus software. One annoying thing that I'm facing and not been able to solve is the duplicated computers GLPI (OCS is ok). Most of the times that I check for new computers, some that were already imported are there again.
I've notice that this mainly happens with Laptops, as they constantly change from networks. By doing it, they change the active network card, and, not always, GLPI will ask to import it again. One note, this also happened with local computers.
I'm using GLPI 0.71.2 and OCS 1.01 for server and agent 4.0.3.2.
What can i do workaround this issue?
Thank you.
Best regards,
Ivo Silva
Offline
hello
do you have the TRACE_DELETED option set to ON in ocsreports ?
Offline
Hi,
Thanks for your reply.
Yes, indeed i have it On. Should I disable it? In the OCS label, it says that it is required on GLPI.
Best regards,
Ivo Silva
Last edited by Ivo.Silva (2008-11-17 13:12:24)
Offline
We have the same problem on our campus, Ivo. We have nearly 1000 devices reporting to our OCS NG application with no duplicates, but on any given day we may have 20-30 "new" computers ready to be imported into GLPI that are nothing but duplicates already in GLPI. Some are laptops, some are desktops. Some have virtual machines, some multi-boot. Some machines show up on a daily basis as another new machine, even...but not all of them act like this. There really is no discovered rationale on our end.
I understand that not everyone uses both GLPI and OCS NG, which is why the convoluted linking table between the two applications exists, but the programmatic design for handling duplicates in the two applications is just so radically different.
I almost wonder if writing a cron task that writes OCS database data (the hardware table, for instance) to the GLPI database (glpi_computers table) is a more simple approach than trying to discover where GLPI is causing these duplicates.
Offline
I work with a OCS database which contains around 30 000 devices. This database is synchronize with GLPI with no problems and no duplicates.
Most problems are coming if your configuration is wrong.
Did you enable fusion in your config GLPI ocs ?
Which criteria do you use for fusion ?
Did you use the plugin mass_ocs_import ? etc...
It's not so difficult to tuning your installation to avoid problem of duplicate.
JMD / Jean-Mathieu Doléans - Glpi-project.org - Association Indepnet
Apportez votre pierre au projet GLPI : Soutenir
Offline
I almost wonder if writing a cron task that writes OCS database data (the hardware table, for instance) to the GLPI database (glpi_computers table) is a more simple approach than trying to discover where GLPI is causing these duplicates.
I don't understand this way. It's exactly what sync do (the better way will probably to handle direct communication between client and GLPI, bypassing the OCS server).
Check if your ID are stable in OCS. A good figure is to count the total number of computers in OCS and the max ID. Also check the history in GLPI for "ocs ID change" event
You can encounter issue if ID change often, more quickly than the cron handle this change.
You also have to check if you don't have "virtual network card" (VPN, vmware, ...) which need to be blacklisted in OCS (directly in the source code for 1.01, from the GUI in 1.02).
We are using mass_ocs_import on a large databse (see my sign), and don't have any issue.
++
Dév. Fedora 29 - PHP 5.6/7.0/7.1/7.2/7.3/7.4 - MariaDB 10.3 - GLPI master
Certifié ITILv3 - RPM pour Fedora, RHEL et CentOS sur https://blog.remirepo.net/
Offline
Hi,
Thank you all for your replies.
JMD:
Could you please explain to me what fusion is and where can i find it? The only fusion option that i've found is for Software, and i'm not using mass_ocs_import pugin.
remi:
In my OCS database the ID file is quit higher than the total number of computers, but this is related to my earlier tests to the solution. In GLPI historical, for all the computers that i've checked, there are no ID Changes......
Any thoughts?
Thanks
Last edited by Ivo.Silva (2008-11-19 13:22:05)
Offline
I don't understand this way. It's exactly what sync do
From what I've seen of the application, the built-in sync doesn't quite do this. In OCS, when it encounters a duplicate, it takes the newer record and deletes the older record. For GLPI, it then syncs the ocs ID to GLPI in a linking table, where GLPI assigns a GLPI ID to the OCS ID, then it builds the GLPI record, using a GLPI-based ID number. But if GLPI ever changes the ID for the record, it has to propagate back to the linking table.
The way I was wondering about uses the same ID for GLPI that is used in OCS, bypassing the linking table altogether. If a change is made in OCS, it will be reflected in GLPI once the cron runs again. I'm sure the idea is full of holes, but I was just brainstorming. Frustrated desperation will drive a man to brainstorming!
(the better way will probably to handle direct communication between client and GLPI, bypassing the OCS server).
I think you're right. OCS for us only inventories the machines and acts as a holding location for that information. It's in the retrieval of this information that GLPI does its harm. If GLPI ever had its own proprietary discovery/reporting agent, it might save us considerable headache.
We're not using the mass_ocs_import at present. With nearly all documentation for GLPI in French, I'm beginning to feel as though I'm operating with a minuscule amount of information. I did run into a nice Firefox translation tool this morning, so I'm going to try and decipher some of the vast French documentation and see if I can get a grasp on some of these suggestions you've made. I'll try to educate myself a little more in those French forums/documentation and try to post thoughtful questions.
Many thanks for those of you non-English speakers doing your best to help us out with documentation. Without you, we'd have no chance.
Offline
Ivo silva : It is not fusion in english. Sorry
In Setup/ocsng/Automatic connection of machines you have to enable :
Activate the automatic link -> yes
Existence criteria of a computer -> and choose good critérias
Warning : The connection automatically connects a GLPI computer with one in OCS. This option is taken into account only by synchronization scripts (IE : the cron job of glpi or the plugin mass ocs import
JMD / Jean-Mathieu Doléans - Glpi-project.org - Association Indepnet
Apportez votre pierre au projet GLPI : Soutenir
Offline
In Setup/ocsng/Automatic connection of machines you have to enable :
Activate the automatic link -> yes
Existence criteria of a computer -> and choose good critérias
Hi JMD
I've done what you say. I'd 'no' on Automatic link.
Regarding the existence criteria, I've choosen:
IP: No
Mac: No
Name of PC: Yes:equal
Serial: Yes
Status: -----
Could you please explain to me what the cron job of GLPI is?
Thanks for you support. Appreciate.
Best regards,
Ivo Silva
Last edited by Ivo.Silva (2008-11-20 11:10:38)
Offline
Hi all,
just to let you know that even with the last changes to the OCSNG and GLPI setup, i'm still getting duplicated computers on GLPI.
This is very annoying as I can't have a perfect control over the computers or software licenses....
Does anyone have any more ideas were to look on?
Thanks for your support.
Best regards,
Offline
Perhaps the critérias you use ti the configuration of automatic link are not suffisant.
The cron job of GLPI : have a look in the wiki documentation (the french section is more updated , use a translator to read)
JMD / Jean-Mathieu Doléans - Glpi-project.org - Association Indepnet
Apportez votre pierre au projet GLPI : Soutenir
Offline
I have the same problems with duplicates.
In OCS (Ver. 1.02 RC3) i have NO duplicates (around 400 computers). In Configuration menu i have:
AUTO_DUPLICATE_LVL:
Hostname: yes
Macaddress: yes
TRACE_DELETED: ON
But each day i have around 5-10 duplicates in GLPI (0.71.3) with configuration (Setup > OCSNG Mode):
Activate the automatic link: Yes
IP: No
Mac: Yes
Name of PC: Yes:equal
Serial: No
Status: -----
All duplicates without changes in comp. configuration, only ID. I don't use any of "cron job of glpi or the plugin mass ocs import" (i don't setup any of them and don't run). But when i open
Tools > OCSNG -> Import new computers, Synchronize computers already imported, Link new OCS computers...
each day i have some new computers there and after some time this lists becomes empty. So i have some questions:
1. Where can i setup cron from GLPI?
2. What difference between "Synchronize computers already imported" and "Link new OCS computers to existing GLPI computers"?
3. Where can i see is computer connected with one in OCS or not, or which duplicated computer connected with one in OCS?
4. Is there any tool to merge duplicates? Which of them i should to delete or to keep?
5. How can i link computer in GLPI with one in OCS?
Offline
Hi there, have also some troubles with duplicates.
We at our company have 560 pcs and day by day we have 6 pcs that keep duplicates.
- 1 is a server (with 3 network adapters)
- 3 workstations that are working with pda's (when pda is connected to pc and u know that pdas have network address)
- 2 laptops
I think that the duplicates appear by some problem in mac address in glpi code.
Any thought about this?
Thanks
Offline
Hi to you all,
I've managed to "solve" this issue by the answers to this post. I've also used Wiki documentation, wich was very helpfull.
I still sometimes get new imports that already exist, but, before every new import, i do a Link new computers to existing, under "Tools--OCSNG".
This isn't still at 100%, but i never get duplicate machines.
I think that this procedure (Link new computers to existing) should be implemented on the New import process.... Just an indea...
Thank you all.
Best regards,
Ivo Silva
Offline
Hi to all,
Any updates regarding these problem?
jmoreno
Offline
Pages: 1