Re: Problem after Nagios restart all host checks completing
Posted: Wed Oct 23, 2013 4:54 am
Hello and thanks for answer 
I don't have any error messages; simply the checks, are performed (I was wrong), but they are not shown correctly on the web interface.
I did a lot of test during last weeks:
- Turned on the nagios.log and I did not find any error messages
- I sniffed the traffic by tcpdump and I saw correctly the nagios request and the host's reponse
- The "checkresults" folder seems correctly working
The configuration, in terms of scheduled check time and many other parameters, was checked many times and it is correct.
I solved the problem splitting the checks on to server and using check_mk in order to watch only one console.
So, I think, is true that the data structure (a list) probably create a perfomance issue two times:
- when a new check result is inserted (maybe an heap could be much better O(log (n)) where n is the number of checks in the list), operation in O(n)
- when the list is parsed in order to show the checks (again the heap data structure should be better, undeerstood as priority queue); the heap perform this operation in costant time O(1) intead of O(n), worst case, of the list.
BTW, splitting the checks on two servers I have great performaces
Bye
Roberto
PS: I hope I have no written too many crap
I don't have any error messages; simply the checks, are performed (I was wrong), but they are not shown correctly on the web interface.
I did a lot of test during last weeks:
- Turned on the nagios.log and I did not find any error messages
- I sniffed the traffic by tcpdump and I saw correctly the nagios request and the host's reponse
- The "checkresults" folder seems correctly working
The configuration, in terms of scheduled check time and many other parameters, was checked many times and it is correct.
I solved the problem splitting the checks on to server and using check_mk in order to watch only one console.
So, I think, is true that the data structure (a list) probably create a perfomance issue two times:
- when a new check result is inserted (maybe an heap could be much better O(log (n)) where n is the number of checks in the list), operation in O(n)
- when the list is parsed in order to show the checks (again the heap data structure should be better, undeerstood as priority queue); the heap perform this operation in costant time O(1) intead of O(n), worst case, of the list.
BTW, splitting the checks on two servers I have great performaces
Bye
Roberto
PS: I hope I have no written too many crap
