Hi! We are using Icinga with NSCA 2.9.1 in single-process mode with delivery to the checkresults directory.
It seems the single-process networking code is not optimal: reading from one TCP connection blocks all other TCP connections until NSCA finished reading all the data from the current TCP connection. Thus, reading data on a slow TCP connection causes delay for all other incoming passive checks. I am a bit afraid switching to multi-process mode due to the high fork rate (sometimes we have up to 100 passive check results per second). So, what do other people use for optimal NSCA performance? single-process or multi-process? Is it worth considering some NSCA replacement? Thanks Klaus _______________________________________________ icinga-users mailing list [email protected] https://lists.icinga.org/mailman/listinfo/icinga-users
