Some clients disappear from web interface but are still present in databse

[Matthieu Jacquart]

It’s not the first time svn version do this to me, but sometimes computer doesn’t appear any more in the web interface. fog.log give me “Middleware::Communication Unknown Response: Invalid Host”, but when I search in database, client exist !
If I try to make a quick registration, client appear in web interface as new client (name = mac address), but if I try to rename it I’ve a message “client already exist” !

It’s problematic, I have 5 or 6 clients who have disappear this morning, I had to suppress them from the database and then recreate them, not convenient at all…

Other problem, I made a group with all clients (around 320), and each time I look at the membership, some clients are not anymore in this group whereas I add them the first time…

Thanks
Matthieu

[Matthieu Jacquart]

Moreover, some clients “lost” services : I check all services and few days after, one or more services are missing…
I suppose it’s a bug due to my database, any idea ?
Can I force fog to check database ?

Thanks

[tmerrick]

I am seeing the same thing on 4493 on Red Hat 6. Random deactivations or even complete deregistration of clients. I had 49 in one classroom and now it is only showing 47. This started happening sometime after 4266 and has been randomly failing since then.

[Tom Elliott]

Do you by chance know exactly what svn or git revision this started occurring?

Does anybody know the particular circumstances to what is happening to cause this? (I think you understand what I mean by this, if not please ask for clarification. – EG. Host reboots, and when it checks in, or everytime it does this sequence?)

[Tom Elliott]

Also, what revision of fog?

[Tom Elliott]

Okay,

I believe I found the issue. This issue would not occur with the new client, it could only occur due to a miss-logic and the Legacy Client.

Please update and let me know when/if you can?

[Matthieu Jacquart]

Ok thanks

For me it happens on computers with new client and on computers with old client.

I’ll update fog and the client and see

[tmerrick]

All of mine are new clients. I will check again next Wednesday. I first saw this during multicast setup under group, but the last few have not been touched for awhile and just seem to disappear. I have a local program and I had to set it to ignore host entries with null names.

[Claude Girard]

After update problem remain.
Rev 4586

• create new group
• include 23 computers in group
• modify for group: active directory, services.

Some computers have no service enabled, some have a few services enabled.
Modifiy again group - services.
some computers dissapeared from group
Delete all computers in group.
Group is empty but in info says 2 members.
Must delete group to restart.

Same with snapins.
I ran all snapins task for a computer.
In snapin tasks list, I see 4 snapins, but computer has only 3. The last has no name.
In database, snapinassoc table link computer to snapin numbered 52, but in snapins table, 52 doesn’t exist.

So how can we verify database consistency ?

Thanks
CG

[Tom Elliott]

@Claude-Girard Thank you for this information. I’ve FINALLY been able to replicate the issue and I believe this to be fixed properly now.

Please update and let me know.

This was a VERY specific set of circumstances, so I much appreciate the “exact” steps to replicate this. I suppose I was lucky to even be able to see the problem first hand finally.

Thank you.

[Matthieu Jacquart]

Sincerely I don’t know, I created groups and put my pc in, I select services and after this sometime it randomly change… But no problems on groups who have been reinstalled long time ago, just bug on newly installed computers.
One possibility is that I’ve got several group with different computers, but I additionally create an group containing all computers ?

[Tom Elliott]

@Matthieu-Jacquart What I’m saying is I was able to verify the issue, though it was very difficult to get to that point. If you do an update (git pull/svn up – however you do it) you should see things functional again.

[Matthieu Jacquart]

I’ve just made the update, but I still have computers who are in database but not in host or group management…

[Claude Girard]

@Matthieu-Jacquart You must clean database before testing again.
Delete bad hosts with phpmyadmin, and make new tests.
I can’t try today but i Will tomorrow

Matthieu tu es Français ?

[Matthieu Jacquart]

Oui français,
Toi aussi je m’en doutais avec ton pseudo
Je teste ça demain, de supprimer les hôtes récalcitrants

[Claude Girard]

@Tom-Elliott IT seems to be ok for me, after several tests with computers in a group with different options in service or active directory or snapins, and after modifiyng them by group, I don’t see anymore strange things

Thank you

[Tom Elliott]

We have finally found the actual issue in hand with this. It was damn near impossible to replicate but @jbob did figure the exact issue…completely by accident. Regardless of how it was found, it is now 100% and truly fixed. It was specific to invalid macs but very hard to explain with understanding. Just know we did find it and we have patched it. As well as a few other bugs and enhancements. Hopefully you all will be much happier.

[Matthieu Jacquart]

I’m sorry but after upgrade, it’s worst than last week…
This morning I modify some groups, and after that at least 10 pc disappear from web interface, but were still in database.
And since svn 4010, I lost around half computer : 170 in web interface, 315 in database.
I’m a little frightened…

[Tom Elliott]

@Matthieu-Jacquart What are the exact things that cause the deletion?

[Matthieu Jacquart]

@Tom-Elliott Chat with Jbob :

I can’t be sure, but my impression is : this morning after the week-end patch, some hosts disappear if I made modifications on their own groups. But since svn 4010, hosts disappear (and very quickly…) from all groups, without modifications made