A exaggerated messages queue

[igarcia]

Hi,

We have a lot of problems with our Nagios XI since 2 month ago, because the nagios xi performance falls down when we applied new configurations.

When we applied new configurations we have a great delay (about 1 hour) from the core and the frontend.

This one isn't the first time we are asking for your support:
https://support.nagios.com/forum/viewto ... 16&t=38029
https://support.nagios.com/forum/viewto ... 16&t=37183

But now the situation is critical for us, because Nagios is the tool for our help desk.

Monitoring the server we can see that messages queue, after a few minutes from a configuration change looks like:

------ Message Queues --------
key msqid owner perms used-bytes messages
0x85000002 8454144 nagios 600 225978368 220682

We believe that the ndo2db module is unable to process its jobs in a parallel way. In order to solve that issue we are considering to deploy the ndo2bd in a multi master cluster architecture. Is it possible? How should be done? Do you have any alternative?

Thanks,

[tmcdonald]

Currently we do not have any documentation for, nor have we tested a multi-master approach in XI. This would not be something we could guarantee or support.

However, looking over your previous posts it appears to updated NDO at one point, but reverted the changes because you saw no benefit. I would recommend upgrading again, because this has worked for people with similar symptoms and is in fact the reason we made those changes. You might not see anything right away, but with that upgrade in-place we can better troubleshoot any additional issues that might be leading to the increased message queue usage, or find further improvements that may need to be made in NDO.

[igarcia]

Hi,

Yes, we deployed the last version looking for a solution, but the messages queues is filled always that we restarted the service.

[root@peslmnagiosxi01 bin]# ./ndo2db -v
./ndo2db: invalid option -- 'v'
NDO2DB 2.1.0b2
Copyright (c) 2009 Nagios Core Development Team and Community Contributors
Copyright (c) 2005-2008 Ethan Galstad
Last Modified: 12-02-2015
-----


Thanks.

[tmcdonald]

Bear in mind that only the last queue (the lowest one in the ipcs -q listing) is in use by NDO, so if there are others that are still filled but the bottom one is almost always at 0, this is healthy.

With this new version in place, what is the status of that queue?

This also might be something we would need a hands-on look at, and for that we would need to move this into a ticket. If you're alright with that, I would suggest emailing our support line with a link to this thread and a descriptive subject. Then we can set up a remote and check things out.

[igarcia]

Hi,

We will test new scenarios (compare unix socket vs remote tcp, apply new configuration with or without alarms, etc.) prior to open the ticket, in order to collect as much information as possible.

Thanks

[mcapra]

Let us know if there are any updates

[igarcia]

We still with the problem


Also, we send a request to the support via mail ([email protected]) the 11th of July without any response.

Can anyone help us?
Thanks.

[tmcdonald]

I see your email in our whitelist now, but if you were not on the list back in July when you emailed us, we would not have received it. Please try sending in once more now that I can confirm you are on the list, and we'll be able to pick up from there.

[igarcia]

Ok,

I understand , the email was send by a one of my team and he is who administrate the nagios server.

Can I change the authorized email to open ticket with the support?

Thanks.

[tmcdonald]

If you email [email protected] they will be able to add or remove authorized emails. We cannot do that from the Support side.