Hi Antony, Thank you for taking time to reply. We are monitoring IOT devices. These devices are in remote locations and behind one-way NAT. We do not have the luxury of unlimited network and network at times can go bad.
*Why passive?* *Because I want the replay logs. Many times the connection b/w master and client breaks and in such scenarios we want the client to keep running the checks and send the check output when the connection re-establishes.* *Is the Master able to connect to the Clients through the NAT connection? The fact that you said above that "the clients (behind NAT) connect to the the master directly" suggests that it cannot (ie: it sounds like one-way NAT), so* *have you made sure that the Master Endpoint definitions in /etc/icinga2/zones.conf on the Clients contains the hostname (or IP address) of the Master, but that the Client Endpoint definitions on the Master does notcontain the hostname or IP address of the Clients?* *No, I dont want the master to connect to client. Yes its a one-way NAT. Yes i have made sure that the master endpoint definitions in zones.conf of the client contain resolvable hostname and client endpoint definitions on the master do not contain the hostname.* *What bandwidth do you have available per client connection?* *I do not want 1 MB per hour. This will cost us 720 MB just for monitoring the IOT device. I want the data to be as much less as possible. * * Please note that i have already optimised all the checks to output as much data as possible.* *That equals around 256 bits per second (if KB means kilobits) or 2kbits per second (if KB means kilobytes).* *That is Kilobytes.* *Is that *really* a significant overhead on your network?* *I am very happy to deploy Icinga2 client on an AWS machine and there we don't care about the data usage at all. But for an IOT device, is it the right tool ? Given i have already given you all the information* *Changing the heartbeat interval from 10s to 100s and log::SetLogPosition interval from 5s to 60s will reduce a significant overhead. But that is hard coded. Can we have it as a configurable entity ?* Good day ! Vikas On Fri, Jan 13, 2017 at 4:30 PM, <icinga-users-requ...@lists.icinga.org> wrote: > Send icinga-users mailing list submissions to > icinga-users@lists.icinga.org > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.icinga.org/mailman/listinfo/icinga-users > or, via email, send a message with subject or body 'help' to > icinga-users-requ...@lists.icinga.org > > You can reach the person managing the list at > icinga-users-ow...@lists.icinga.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of icinga-users digest..." > > > Today's Topics: > > 1. Icinga2 reduce network data usage (Vikas Tiwari) > 2. Re: Icinga2 reduce network data usage (Antony Stone) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Thu, 12 Jan 2017 18:43:48 +0530 > From: Vikas Tiwari <vikask...@gmail.com> > To: icinga-users@lists.icinga.org > Subject: [icinga-users] Icinga2 reduce network data usage > Message-ID: > <CAEVQhqrWHmYCxsXGpTVP+fRiOucKsrHFJD3P2pOd4RSsE40weg@ > mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > Hi, > > I am testing Icinga2 as a monitoring tool for our infrastructure which will > monitor close to 3000 nodes. > > I have a top down config sync. There is a single master and and there are > no fail overs(for now). The clients(behind NAT) connect to the master > directly. > > I have configured around 20 passive checks on each client. What i have > observed is that the heartbeats from both the ends i.e the master and the > clients are very aggressive which ping every 10 seconds. > > Further the JSONRPC pings are as frequent as 5 seconds, there is also a lot > of other data exchanged between the client and the master. This all leads > to high consumption of network data, Given we have limited data on the > client machines we cannot afford to spend this much data just for > monitoring. > > Currently, per hour 800-900 KB is consumed.Out of this only around 300KB > seems to be the check output. Check intervals are 6 minutes. > > The hearbeat and other intervals are hard coded in the source code itself > which will be my last option. > > Is there any other way i can reduce the network data consumption. ? > > Vikas Tiwari > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: <http://lists.icinga.org/pipermail/icinga-users/ > attachments/20170112/14ecfd34/attachment-0001.html> > > ------------------------------ > > Message: 2 > Date: Thu, 12 Jan 2017 14:58:41 +0100 > From: Antony Stone <antony.st...@icinga.open.source.it> > To: "Icinga User's Corner" <icinga-users@lists.icinga.org> > Subject: Re: [icinga-users] Icinga2 reduce network data usage > Message-ID: <201701121458.42028.antony.st...@icinga.open.source.it> > Content-Type: Text/Plain; charset="iso-8859-15" > > On Thursday 12 January 2017 at 14:13:48, Vikas Tiwari wrote: > > > Hi, > > > > I am testing Icinga2 as a monitoring tool for our infrastructure which > will > > monitor close to 3000 nodes. > > > > I have a top down config sync. There is a single master and and there are > > no fail overs(for now). The clients(behind NAT) connect to the master > > directly. > > > > I have configured around 20 passive checks on each client. > > Why passive? > > > What i have observed is that the heartbeats from both the ends i.e the > > master and the clients are very aggressive which ping every 10 seconds. > > Is the Master able to connect to the Clients through the NAT connection? > The > fact that you said above that "the clients (behind NAT) connect to the the > master directly" suggests that it cannot (ie: it sounds like one-way NAT), > so > have you made sure that the Master Endpoint definitions in > /etc/icinga2/zones.conf on the Clients contains the hostname (or IP > address) > of the Master, but that the Client Endpoint definitions on the Master does > not > contain the hostname or IP address of the Clients? > > > Further the JSONRPC pings are as frequent as 5 seconds, there is also a > lot > > of other data exchanged between the client and the master. This all leads > > to high consumption of network data, Given we have limited data on the > > client machines we cannot afford to spend this much data just for > > monitoring. > > What bandwidth do you have available per client connection? > > > Currently, per hour 800-900 KB is consumed. > > That equals around 256 bits per second (if KB means kilobits) or 2kbits per > second (if KB means kilobytes). > > Is that *really* a significant overhead on your network? > > > Antony. > > -- > I want to build a machine that will be proud of me. > > - Danny Hillis, creator of The Connection Machine > > Please reply to the > list; > please *don't* CC > me. > > > ------------------------------ > > Subject: Digest Footer > > _______________________________________________ > icinga-users mailing list > icinga-users@lists.icinga.org > https://lists.icinga.org/mailman/listinfo/icinga-users > > > ------------------------------ > > End of icinga-users Digest, Vol 37, Issue 9 > ******************************************* >
_______________________________________________ icinga-users mailing list icinga-users@lists.icinga.org https://lists.icinga.org/mailman/listinfo/icinga-users
