Monitoring-Portal

the blocking behaviour of the ndo is a very common problem. if the ndo2db-daemon encoutners a problem it causes the ndomod writing to the socket to go over to blocking mode which recently causes nagios to stay like a dead process but still running but not doing any checks neither by process nor by request on the web. we've encountered that on a weekend therefore a small perlscript which checks the actuality of status.dat (where all the checks are logged into) has been applied on another server to see if there's a problem with nagios - kind of strange but well well.

removing some kind of blocking by trimming the database regularly could be a solution. database trimming works like this:

* given several interval parameters for the most common used tables in ndo2db.cfg
* at ndo2db startup, those tables will be checked and by a simple "delete from bla where time < (now - interval)" query. this could lead into quite a long deletion period, seen that on oracle where ndo2db simply died deleting from 3,7 mio rows
* running ndo2db, every 60 seconds a check for old entries is being performed, likewise quite the same delete statement to wait for

concerning the amount of data - ndo2db saves nearly everything from nagios, even event logs (timed_events) which is kind of annoying if you do not need the data. for historical purposes and analysis of services and their up/downtime of the past year e.g. it's a good thing to have (even though mysql shouldn't be the preferred rdbm that way, also the db scheme is way too normalized).
if you desire not to have several data, you may change the data_processing_options in the config to an appropriate value (i.e. kick timed_events and so on). several addons still require only the configs and some live data for them to work with ndoutils.-

first of all, i've become one of the maintainers for ndoutils oracle which is a port of ndoutils and using mysql, just for oracle. as you said, since the development of ndoutils quietly stopped and the codebase wasn't that good i got a bit more deeper into the code. just as icinga was announced as a fork and the code has been patched a bit better with ndoutils -> idoutils. primary goal of ndoutils oracle was, to commit changes to the upstream which wasn't possible quite a long time. for that reason i have taken the chances to commit knowledge to icinga and idoutils. currently i am working on porting and improving queries and housekeeping for mysql, and getting postgres and also oracle to work with idoutils and icinga.

the blocking behaviour is a well known fact it has been discussed on the mailinglist in different threads but since development and support was not there, nothing has changed. we have discussed it here recently (but in german). there are several lines in the code where to start debugging and build a better "buffer" to catch on that bug. let's see if we can catch up on that in icinga and improve the idoutils. if you may have any further questions just come back here. imho the nagios-users mailinglist is quite good, but not that helpful and searching for a solutions is sometimes kind of weird.

as i pointed out for your problem, try the database_trimming_options and in first line, decide with data_processing_options what you really need in your database. if your not interested in historical data, just don't allow the ndo to get it and write it to the database. in ndomod.h you will find the defines with values - add them alltogether what you need and that's the value to set in the configs (also something not quite userfriendly and probably to be changed if the uersrs want that...). default value is btw -1, meaining everything.

ethan galstad (the so-called "godfather" of nagios) recently searched for ndoutils developers since it came to his mind that there was a lack of developers. hendrik and me tried to apply in case of merging fixes for icinga/idoutils to actual ndoutils. what happened next that hendrik got into the team while myself is a bit unknown to the nagios "scene" since i was not that active on the mailinglist and so on (i understand that aspect but concerning my programming skills well.. no).

so right now, 1.4b8 has been released which is a patch compilation since the last release (i've been using and testing release only from cvs). maybe development grows up a bit more again and several fixes wil be applied.

which brings us to icinga - the fork was applied as a consequence of ethan galstad being the only nagios developer and allowing/testing comitted patches. that is not just that the community is awaiting. so there was the split but the keep in mind fact is still to rename code without any nagios trademark to something new and then work and improve the code. but right this way with a broader bandwiddth of people working together and also introducing new features - this is what icinga is.

as a matter of fact, idoutils .e.g should work with nagios. but with icinga, it is currently part of the main install of icinga. meaning, a module next to the core which will be improved in many ways. on the mailinglist there was a discussion to completly delete it and code something new (which merlin from op5 tries to be). but, main goals are to bringt more rdbms to work with idoutils, improve buffering/housekeeping and so on. heavily focused on doing things right which ndoutils couldn't fix. as an example, if you look at the actual ndoutils code and before you will notice that there should have been an implementation for postgres. but in fact only delete and insert statements would have fit, everything else is not yet implemented or pure "bullshit".

Probably a Summertime-Problem?
See:
http://www.nagios-portal.org/wbb/index.p…d&threadID=4930