-
Delay of the Pandora servers
Hello !
In the servers tab in Pandora, what is the delay time indicated and the number of modules waiting and late? Where can it come from and can it affect the operation of Pandora (processing SNMP traps or triggering alerts)? How can we correct this delay?
I attach a screenshot of the page in question.
Thanks in advance,
have a nice day,
-
4 Replies
-
Hello,
The Lag indicator in the Server’s View, is just an aproximate indicator which tells you that there are some modules queued to be processed yet.
This is not something to be worried about if there are no big indicators of queueing modules that might be caused by some malfunction of the server which is showing that indicator.
In your case there are some dataserver modules that are being queued, and from what we are able to see in the picture you have more than 14000 modules running on only 4 threads.
In order to improve this lag you’re experiencing in the dataserver you will need to increase the threads from 6 up to 8 (8 will be recommended in this case)
For the Plugins server the Lag tends to be more normal, as some plugins take quite some time to end up executing which causes some Lag.
Probably there is some Plugin with some timeout configured which may be causing this Lags and queues, even I wouldn’t worry so much about this ones for now, you can check the plugins and see if there is any of those plugins with a big timeout configured, also be aware that there is a parameter inside pandora_server.conf file called plugin_timeout which have priority over the Max Timeout set in each individual plugin configurations inside the Console.
Kind regards,
Sergio B.
-
Hello,
Thank you for your answer.
We are going to do what is necessary to resize our servers and increase the number of threads allocated to the data server.
We sometimes have a lot of delay (30 minutes) and a lot of pending modules (more than 60) on the data server, can this cause an alert pattern to be triggered several days after the condition that is supposed to trigger it has occurred?
Also we sometimes have delays and delayed modules on the SNMP server, does this mean that some of the traps received are not processed? Is there a solution to solve this delay?
Thanks again,
Best regards
-
Hello,
The SNMP server is not in charge of the TRAPS, in this case the server in charge of the TRAPS is the snmptrap server which is configured in the server config file as snmpconsole 1 you can also improve the threads of this server with the parameter snmpconsole_threads and set this to the number of threads you need.
Please bear in mind that you cannot have more than 40 threads assigned, if you configure more than 40 threads the server will start malfunctioning and you will need to reduce threads in this case.
The SNMP server can have some delays due to latency in the network, when the modules take more time than desired to get the information you have the parameter snmp_timeout inside the pandora_server.conf file in order to cut the request in case its generating unwanted delays in the server.
Kind regards,
Sergio B.
-
Hello,
Thank you for your answer, is there a way to calculate the number of threads assignable to the different Pandora services according to the capacity of our servers (CPU/number of cores)?
To check that we have not exceeded the maximum number of threads, is it enough to count the number of threads associated with the Pandora services in the pandora_server.conf file or are there others to take into account?
Thank you in advance for your answer,
best regards
