nagios-db-devel

[Nagios-db-devel] retention.dat

[Matthew K. <mk...@ma...>]

I've been working on the last bits of my MySQL neb module have just
noticed a rather basic issue for both modules that hasn't cropped up
till now. 

It seems that without the retention.dat file nagios doesn't feed the big
lump of host/service data into the module on startup. Now logically this
makes sense but causes a bit of a problem when there's hostgroup
memberships to insert into the db that depend on those entries getting
loaded at startup. This also messes up adding a host to nagios and
restarting it as again the host won't be in the db until its first
checks are finished breaking the hostgroup data.

As for possible solutions, looping through the lists of hosts and
services in processStart and registering them in the db is a
possibility. But if the retention file already exists your doing twice
the work since nagios will still send all the broker calls, and I'm not
sure of a way to detect this in processStatus to skip those entries. 

Another (and I think better) option would be to submit a patch request
to Ethan to have Nagios stop sending event broker updates when reading
retention.dat and simply have it loop through both host and service
lists after reading it (whether it exists or not) and send all the data
at every startup.

Thoughts? Am I missing something obvious...
-- 
Matthew Kent <mk...@ma...>
http://magoazul.com

Re: [Nagios-db-devel] retention.dat

[Ben <be...@si...>]

Before I can speak to the solutions, I don't understand the problem.

1. Why would retention.dat be missing?
2. Hostgroup membership is determined by walking the hostgroup config 
object in memory.
3. It's true that if nagios is told to re-read its config, it might not 
re-initialize the NEB, which might be a problem for us. On the other 
hand, it might not, because we re-read the config stuff after 
initialization but before steady state.
4. Every time I restart Nagios, I get an imperial boatload of updates. 
Then again, I probably have a retention.dat file.

On Jan 3, 2005, at 9:40 PM, Matthew Kent wrote:

> I've been working on the last bits of my MySQL neb module have just
> noticed a rather basic issue for both modules that hasn't cropped up
> till now.
>
> It seems that without the retention.dat file nagios doesn't feed the 
> big
> lump of host/service data into the module on startup. Now logically 
> this
> makes sense but causes a bit of a problem when there's hostgroup
> memberships to insert into the db that depend on those entries getting
> loaded at startup. This also messes up adding a host to nagios and
> restarting it as again the host won't be in the db until its first
> checks are finished breaking the hostgroup data.
>
> As for possible solutions, looping through the lists of hosts and
> services in processStart and registering them in the db is a
> possibility. But if the retention file already exists your doing twice
> the work since nagios will still send all the broker calls, and I'm not
> sure of a way to detect this in processStatus to skip those entries.
>
> Another (and I think better) option would be to submit a patch request
> to Ethan to have Nagios stop sending event broker updates when reading
> retention.dat and simply have it loop through both host and service
> lists after reading it (whether it exists or not) and send all the data
> at every startup.
>
> Thoughts? Am I missing something obvious...
> -- 
> Matthew Kent <mk...@ma...>
> http://magoazul.com
>
>
>
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Nagios-db-devel mailing list
> Nag...@li...
> https://lists.sourceforge.net/lists/listinfo/nagios-db-devel

Re: [Nagios-db-devel] retention.dat

[Matthew K. <mk...@ma...>]

On Tue, 2005-01-04 at 15:34, Ben wrote:
> Before I can speak to the solutions, I don't understand the problem.
> 
> 1. Why would retention.dat be missing?

On a new install, server move or after wiping cached data in var/.

> 2. Hostgroup membership is determined by walking the hostgroup config 
> object in memory.
> 3. It's true that if nagios is told to re-read its config, it might not 
> re-initialize the NEB, which might be a problem for us. On the other 
> hand, it might not, because we re-read the config stuff after 
> initialization but before steady state.

Sorry for the confusion here, I mean if you add a new host or service to
the config and restart nagios the corresponding hostgroup_membership
insert will fail for this new host or service as it will not be
registered in the host/service tables until the first check is made. Of
course you could wait x minutes until the retention file is saved and
restart nagios again so the config data will be right, but that's not
much of a solution.

> 4. Every time I restart Nagios, I get an imperial boatload of updates. 
> Then again, I probably have a retention.dat file.
> 

Yeah wipe the retention file and try it to see what I mean :)

- Matt

> On Jan 3, 2005, at 9:40 PM, Matthew Kent wrote:
> 
> > I've been working on the last bits of my MySQL neb module have just
> > noticed a rather basic issue for both modules that hasn't cropped up
> > till now.
> >
> > It seems that without the retention.dat file nagios doesn't feed the 
> > big
> > lump of host/service data into the module on startup. Now logically 
> > this
> > makes sense but causes a bit of a problem when there's hostgroup
> > memberships to insert into the db that depend on those entries getting
> > loaded at startup. This also messes up adding a host to nagios and
> > restarting it as again the host won't be in the db until its first
> > checks are finished breaking the hostgroup data.
> >
> > As for possible solutions, looping through the lists of hosts and
> > services in processStart and registering them in the db is a
> > possibility. But if the retention file already exists your doing twice
> > the work since nagios will still send all the broker calls, and I'm not
> > sure of a way to detect this in processStatus to skip those entries.
> >
> > Another (and I think better) option would be to submit a patch request
> > to Ethan to have Nagios stop sending event broker updates when reading
> > retention.dat and simply have it loop through both host and service
> > lists after reading it (whether it exists or not) and send all the data
> > at every startup.
> >
> > Thoughts? Am I missing something obvious...
> > -- 
> > Matthew Kent <mk...@ma...>
> > http://magoazul.com
> >
> >
> >
> > -------------------------------------------------------
> > The SF.Net email is sponsored by: Beat the post-holiday blues
> > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > _______________________________________________
> > Nagios-db-devel mailing list
> > Nag...@li...
> > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> 
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Nagios-db-devel mailing list
> Nag...@li...
> https://lists.sourceforge.net/lists/listinfo/nagios-db-devel

Re: [Nagios-db-devel] retention.dat

[Ben <be...@si...>]

Ahhhhh..... I understand now. I *knew* I should have checked for insert 
status. 

Now, as for the two solutions you proposed, they seem the same, except 
that the latter includes modifying nagios to not send updates when parsing 
retention.dat data. Correct?

On Tue, 4 Jan 2005, Matthew Kent wrote:

> On Tue, 2005-01-04 at 15:34, Ben wrote:
> > Before I can speak to the solutions, I don't understand the problem.
> > 
> > 1. Why would retention.dat be missing?
> 
> On a new install, server move or after wiping cached data in var/.
> 
> > 2. Hostgroup membership is determined by walking the hostgroup config 
> > object in memory.
> > 3. It's true that if nagios is told to re-read its config, it might not 
> > re-initialize the NEB, which might be a problem for us. On the other 
> > hand, it might not, because we re-read the config stuff after 
> > initialization but before steady state.
> 
> Sorry for the confusion here, I mean if you add a new host or service to
> the config and restart nagios the corresponding hostgroup_membership
> insert will fail for this new host or service as it will not be
> registered in the host/service tables until the first check is made. Of
> course you could wait x minutes until the retention file is saved and
> restart nagios again so the config data will be right, but that's not
> much of a solution.
> 
> > 4. Every time I restart Nagios, I get an imperial boatload of updates. 
> > Then again, I probably have a retention.dat file.
> > 
> 
> Yeah wipe the retention file and try it to see what I mean :)
> 
> - Matt
> 
> > On Jan 3, 2005, at 9:40 PM, Matthew Kent wrote:
> > 
> > > I've been working on the last bits of my MySQL neb module have just
> > > noticed a rather basic issue for both modules that hasn't cropped up
> > > till now.
> > >
> > > It seems that without the retention.dat file nagios doesn't feed the 
> > > big
> > > lump of host/service data into the module on startup. Now logically 
> > > this
> > > makes sense but causes a bit of a problem when there's hostgroup
> > > memberships to insert into the db that depend on those entries getting
> > > loaded at startup. This also messes up adding a host to nagios and
> > > restarting it as again the host won't be in the db until its first
> > > checks are finished breaking the hostgroup data.
> > >
> > > As for possible solutions, looping through the lists of hosts and
> > > services in processStart and registering them in the db is a
> > > possibility. But if the retention file already exists your doing twice
> > > the work since nagios will still send all the broker calls, and I'm not
> > > sure of a way to detect this in processStatus to skip those entries.
> > >
> > > Another (and I think better) option would be to submit a patch request
> > > to Ethan to have Nagios stop sending event broker updates when reading
> > > retention.dat and simply have it loop through both host and service
> > > lists after reading it (whether it exists or not) and send all the data
> > > at every startup.
> > >
> > > Thoughts? Am I missing something obvious...
> > > -- 
> > > Matthew Kent <mk...@ma...>
> > > http://magoazul.com
> > >
> > >
> > >
> > > -------------------------------------------------------
> > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > _______________________________________________
> > > Nagios-db-devel mailing list
> > > Nag...@li...
> > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > 
> > 
> > 
> > -------------------------------------------------------
> > The SF.Net email is sponsored by: Beat the post-holiday blues
> > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > _______________________________________________
> > Nagios-db-devel mailing list
> > Nag...@li...
> > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> 
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Nagios-db-devel mailing list
> Nag...@li...
> https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> 

Re: [Nagios-db-devel] retention.dat

[Matthew K. <mk...@ma...>]

On Tue, 2005-01-04 at 17:14, Ben wrote:
> Ahhhhh..... I understand now. I *knew* I should have checked for insert 
> status. 
> 
> Now, as for the two solutions you proposed, they seem the same, except 
> that the latter includes modifying nagios to not send updates when parsing 
> retention.dat data. Correct?
> 

Almost, my idea was to either solve it only the module in a hackish way
or just patch nagios directly. 

I do have a better idea though. If I recall from my testing processStart
got triggered after all the data from the retention file was loaded
(triggering that big wad of updates). So if the neb register callbacks
for host and status data were moved to the end of processStart, you
could simply walk the host and service list and the beginning of the
function and pass the data to processStatus to register. Basically
ignore the first round of data from nagios loading the retention file
and do it ourselves.

- Matt

> On Tue, 4 Jan 2005, Matthew Kent wrote:
> 
> > On Tue, 2005-01-04 at 15:34, Ben wrote:
> > > Before I can speak to the solutions, I don't understand the problem.
> > > 
> > > 1. Why would retention.dat be missing?
> > 
> > On a new install, server move or after wiping cached data in var/.
> > 
> > > 2. Hostgroup membership is determined by walking the hostgroup config 
> > > object in memory.
> > > 3. It's true that if nagios is told to re-read its config, it might not 
> > > re-initialize the NEB, which might be a problem for us. On the other 
> > > hand, it might not, because we re-read the config stuff after 
> > > initialization but before steady state.
> > 
> > Sorry for the confusion here, I mean if you add a new host or service to
> > the config and restart nagios the corresponding hostgroup_membership
> > insert will fail for this new host or service as it will not be
> > registered in the host/service tables until the first check is made. Of
> > course you could wait x minutes until the retention file is saved and
> > restart nagios again so the config data will be right, but that's not
> > much of a solution.
> > 
> > > 4. Every time I restart Nagios, I get an imperial boatload of updates. 
> > > Then again, I probably have a retention.dat file.
> > > 
> > 
> > Yeah wipe the retention file and try it to see what I mean :)
> > 
> > - Matt
> > 
> > > On Jan 3, 2005, at 9:40 PM, Matthew Kent wrote:
> > > 
> > > > I've been working on the last bits of my MySQL neb module have just
> > > > noticed a rather basic issue for both modules that hasn't cropped up
> > > > till now.
> > > >
> > > > It seems that without the retention.dat file nagios doesn't feed the 
> > > > big
> > > > lump of host/service data into the module on startup. Now logically 
> > > > this
> > > > makes sense but causes a bit of a problem when there's hostgroup
> > > > memberships to insert into the db that depend on those entries getting
> > > > loaded at startup. This also messes up adding a host to nagios and
> > > > restarting it as again the host won't be in the db until its first
> > > > checks are finished breaking the hostgroup data.
> > > >
> > > > As for possible solutions, looping through the lists of hosts and
> > > > services in processStart and registering them in the db is a
> > > > possibility. But if the retention file already exists your doing twice
> > > > the work since nagios will still send all the broker calls, and I'm not
> > > > sure of a way to detect this in processStatus to skip those entries.
> > > >
> > > > Another (and I think better) option would be to submit a patch request
> > > > to Ethan to have Nagios stop sending event broker updates when reading
> > > > retention.dat and simply have it loop through both host and service
> > > > lists after reading it (whether it exists or not) and send all the data
> > > > at every startup.
> > > >
> > > > Thoughts? Am I missing something obvious...
> > > > -- 
> > > > Matthew Kent <mk...@ma...>
> > > > http://magoazul.com
> > > >
> > > >
> > > >
> > > > -------------------------------------------------------
> > > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > > _______________________________________________
> > > > Nagios-db-devel mailing list
> > > > Nag...@li...
> > > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > > 
> > > 
> > > 
> > > -------------------------------------------------------
> > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > _______________________________________________
> > > Nagios-db-devel mailing list
> > > Nag...@li...
> > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > 
> > 
> > 
> > -------------------------------------------------------
> > The SF.Net email is sponsored by: Beat the post-holiday blues
> > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > _______________________________________________
> > Nagios-db-devel mailing list
> > Nag...@li...
> > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > 
> 
> 
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Nagios-db-devel mailing list
> Nag...@li...
> https://lists.sourceforge.net/lists/listinfo/nagios-db-devel

Re: [Nagios-db-devel] retention.dat

[Ben <be...@si...>]

Well, what I understand sounds like a good solution. Let's see if I 
understand it right....

1. On NEB init, start up a database connection and register for 
processStart.

2. On processStart, parse all the hosts. If they aren't in the database
yet, add them. Refresh all the hostgroups. Register for status events.

3. Steadystate remains unchanged.

Right?

Actually, we're already parsing all the hosts in order to mark the ones 
that are currently configured, so this shouldn't be so bad.

On Tue, 4 Jan 2005, Matthew Kent wrote:

> On Tue, 2005-01-04 at 17:14, Ben wrote:
> > Ahhhhh..... I understand now. I *knew* I should have checked for insert 
> > status. 
> > 
> > Now, as for the two solutions you proposed, they seem the same, except 
> > that the latter includes modifying nagios to not send updates when parsing 
> > retention.dat data. Correct?
> > 
> 
> Almost, my idea was to either solve it only the module in a hackish way
> or just patch nagios directly. 
> 
> I do have a better idea though. If I recall from my testing processStart
> got triggered after all the data from the retention file was loaded
> (triggering that big wad of updates). So if the neb register callbacks
> for host and status data were moved to the end of processStart, you
> could simply walk the host and service list and the beginning of the
> function and pass the data to processStatus to register. Basically
> ignore the first round of data from nagios loading the retention file
> and do it ourselves.
> 
> - Matt
> 
> > On Tue, 4 Jan 2005, Matthew Kent wrote:
> > 
> > > On Tue, 2005-01-04 at 15:34, Ben wrote:
> > > > Before I can speak to the solutions, I don't understand the problem.
> > > > 
> > > > 1. Why would retention.dat be missing?
> > > 
> > > On a new install, server move or after wiping cached data in var/.
> > > 
> > > > 2. Hostgroup membership is determined by walking the hostgroup config 
> > > > object in memory.
> > > > 3. It's true that if nagios is told to re-read its config, it might not 
> > > > re-initialize the NEB, which might be a problem for us. On the other 
> > > > hand, it might not, because we re-read the config stuff after 
> > > > initialization but before steady state.
> > > 
> > > Sorry for the confusion here, I mean if you add a new host or service to
> > > the config and restart nagios the corresponding hostgroup_membership
> > > insert will fail for this new host or service as it will not be
> > > registered in the host/service tables until the first check is made. Of
> > > course you could wait x minutes until the retention file is saved and
> > > restart nagios again so the config data will be right, but that's not
> > > much of a solution.
> > > 
> > > > 4. Every time I restart Nagios, I get an imperial boatload of updates. 
> > > > Then again, I probably have a retention.dat file.
> > > > 
> > > 
> > > Yeah wipe the retention file and try it to see what I mean :)
> > > 
> > > - Matt
> > > 
> > > > On Jan 3, 2005, at 9:40 PM, Matthew Kent wrote:
> > > > 
> > > > > I've been working on the last bits of my MySQL neb module have just
> > > > > noticed a rather basic issue for both modules that hasn't cropped up
> > > > > till now.
> > > > >
> > > > > It seems that without the retention.dat file nagios doesn't feed the 
> > > > > big
> > > > > lump of host/service data into the module on startup. Now logically 
> > > > > this
> > > > > makes sense but causes a bit of a problem when there's hostgroup
> > > > > memberships to insert into the db that depend on those entries getting
> > > > > loaded at startup. This also messes up adding a host to nagios and
> > > > > restarting it as again the host won't be in the db until its first
> > > > > checks are finished breaking the hostgroup data.
> > > > >
> > > > > As for possible solutions, looping through the lists of hosts and
> > > > > services in processStart and registering them in the db is a
> > > > > possibility. But if the retention file already exists your doing twice
> > > > > the work since nagios will still send all the broker calls, and I'm not
> > > > > sure of a way to detect this in processStatus to skip those entries.
> > > > >
> > > > > Another (and I think better) option would be to submit a patch request
> > > > > to Ethan to have Nagios stop sending event broker updates when reading
> > > > > retention.dat and simply have it loop through both host and service
> > > > > lists after reading it (whether it exists or not) and send all the data
> > > > > at every startup.
> > > > >
> > > > > Thoughts? Am I missing something obvious...
> > > > > -- 
> > > > > Matthew Kent <mk...@ma...>
> > > > > http://magoazul.com
> > > > >
> > > > >
> > > > >
> > > > > -------------------------------------------------------
> > > > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > > > _______________________________________________
> > > > > Nagios-db-devel mailing list
> > > > > Nag...@li...
> > > > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > > > 
> > > > 
> > > > 
> > > > -------------------------------------------------------
> > > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > > _______________________________________________
> > > > Nagios-db-devel mailing list
> > > > Nag...@li...
> > > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > > 
> > > 
> > > 
> > > -------------------------------------------------------
> > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > _______________________________________________
> > > Nagios-db-devel mailing list
> > > Nag...@li...
> > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > > 
> > 
> > 
> > 
> > 
> > -------------------------------------------------------
> > The SF.Net email is sponsored by: Beat the post-holiday blues
> > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > _______________________________________________
> > Nagios-db-devel mailing list
> > Nag...@li...
> > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> 
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Nagios-db-devel mailing list
> Nag...@li...
> https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> 

Re: [Nagios-db-devel] retention.dat

[Matthew K. <mk...@ma...>]

On Tue, 2005-01-04 at 18:06, Ben wrote:
> Well, what I understand sounds like a good solution. Let's see if I 
> understand it right....
> 
> 1. On NEB init, start up a database connection and register for 
> processStart.
> 
> 2. On processStart, parse all the hosts. If they aren't in the database
> yet, add them. Refresh all the hostgroups. Register for status events.
> 
> 3. Steadystate remains unchanged.
> 
> Right?
> 

Yeah that sounds right. Hope that works. Sorry I can't test first to
confirm, got a few other things to do this afternoon.

- Matt

> Actually, we're already parsing all the hosts in order to mark the ones 
> that are currently configured, so this shouldn't be so bad.
> 
> On Tue, 4 Jan 2005, Matthew Kent wrote:
> 
> > On Tue, 2005-01-04 at 17:14, Ben wrote:
> > > Ahhhhh..... I understand now. I *knew* I should have checked for insert 
> > > status. 
> > > 
> > > Now, as for the two solutions you proposed, they seem the same, except 
> > > that the latter includes modifying nagios to not send updates when parsing 
> > > retention.dat data. Correct?
> > > 
> > 
> > Almost, my idea was to either solve it only the module in a hackish way
> > or just patch nagios directly. 
> > 
> > I do have a better idea though. If I recall from my testing processStart
> > got triggered after all the data from the retention file was loaded
> > (triggering that big wad of updates). So if the neb register callbacks
> > for host and status data were moved to the end of processStart, you
> > could simply walk the host and service list and the beginning of the
> > function and pass the data to processStatus to register. Basically
> > ignore the first round of data from nagios loading the retention file
> > and do it ourselves.
> > 
> > - Matt
> > 
> > > On Tue, 4 Jan 2005, Matthew Kent wrote:
> > > 
> > > > On Tue, 2005-01-04 at 15:34, Ben wrote:
> > > > > Before I can speak to the solutions, I don't understand the problem.
> > > > > 
> > > > > 1. Why would retention.dat be missing?
> > > > 
> > > > On a new install, server move or after wiping cached data in var/.
> > > > 
> > > > > 2. Hostgroup membership is determined by walking the hostgroup config 
> > > > > object in memory.
> > > > > 3. It's true that if nagios is told to re-read its config, it might not 
> > > > > re-initialize the NEB, which might be a problem for us. On the other 
> > > > > hand, it might not, because we re-read the config stuff after 
> > > > > initialization but before steady state.
> > > > 
> > > > Sorry for the confusion here, I mean if you add a new host or service to
> > > > the config and restart nagios the corresponding hostgroup_membership
> > > > insert will fail for this new host or service as it will not be
> > > > registered in the host/service tables until the first check is made. Of
> > > > course you could wait x minutes until the retention file is saved and
> > > > restart nagios again so the config data will be right, but that's not
> > > > much of a solution.
> > > > 
> > > > > 4. Every time I restart Nagios, I get an imperial boatload of updates. 
> > > > > Then again, I probably have a retention.dat file.
> > > > > 
> > > > 
> > > > Yeah wipe the retention file and try it to see what I mean :)
> > > > 
> > > > - Matt
> > > > 
> > > > > On Jan 3, 2005, at 9:40 PM, Matthew Kent wrote:
> > > > > 
> > > > > > I've been working on the last bits of my MySQL neb module have just
> > > > > > noticed a rather basic issue for both modules that hasn't cropped up
> > > > > > till now.
> > > > > >
> > > > > > It seems that without the retention.dat file nagios doesn't feed the 
> > > > > > big
> > > > > > lump of host/service data into the module on startup. Now logically 
> > > > > > this
> > > > > > makes sense but causes a bit of a problem when there's hostgroup
> > > > > > memberships to insert into the db that depend on those entries getting
> > > > > > loaded at startup. This also messes up adding a host to nagios and
> > > > > > restarting it as again the host won't be in the db until its first
> > > > > > checks are finished breaking the hostgroup data.
> > > > > >
> > > > > > As for possible solutions, looping through the lists of hosts and
> > > > > > services in processStart and registering them in the db is a
> > > > > > possibility. But if the retention file already exists your doing twice
> > > > > > the work since nagios will still send all the broker calls, and I'm not
> > > > > > sure of a way to detect this in processStatus to skip those entries.
> > > > > >
> > > > > > Another (and I think better) option would be to submit a patch request
> > > > > > to Ethan to have Nagios stop sending event broker updates when reading
> > > > > > retention.dat and simply have it loop through both host and service
> > > > > > lists after reading it (whether it exists or not) and send all the data
> > > > > > at every startup.
> > > > > >
> > > > > > Thoughts? Am I missing something obvious...
> > > > > > -- 
> > > > > > Matthew Kent <mk...@ma...>
> > > > > > http://magoazul.com
> > > > > >
> > > > > >
> > > > > >
> > > > > > -------------------------------------------------------
> > > > > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > > > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > > > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > > > > _______________________________________________
> > > > > > Nagios-db-devel mailing list
> > > > > > Nag...@li...
> > > > > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > > > > 
> > > > > 
> > > > > 
> > > > > -------------------------------------------------------
> > > > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > > > _______________________________________________
> > > > > Nagios-db-devel mailing list
> > > > > Nag...@li...
> > > > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > > > 
> > > > 
> > > > 
> > > > -------------------------------------------------------
> > > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > > _______________________________________________
> > > > Nagios-db-devel mailing list
> > > > Nag...@li...
> > > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > > > 
> > > 
> > > 
> > > 
> > > 
> > > -------------------------------------------------------
> > > The SF.Net email is sponsored by: Beat the post-holiday blues
> > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > > _______________________________________________
> > > Nagios-db-devel mailing list
> > > Nag...@li...
> > > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > 
> > 
> > 
> > -------------------------------------------------------
> > The SF.Net email is sponsored by: Beat the post-holiday blues
> > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> > _______________________________________________
> > Nagios-db-devel mailing list
> > Nag...@li...
> > https://lists.sourceforge.net/lists/listinfo/nagios-db-devel
> > 
> 
> 
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Nagios-db-devel mailing list
> Nag...@li...
> https://lists.sourceforge.net/lists/listinfo/nagios-db-devel

Re: [Nagios-db-devel] retention.dat

[Ben <be...@si...>]

On Tue, 4 Jan 2005, Matthew Kent wrote:

> 
> Yeah that sounds right. Hope that works. Sorry I can't test first to
> confirm, got a few other things to do this afternoon.
> 
> - Matt

Not a problem. I just checked in stuff that should do this. At least, it 
seems to work on my side with minimal testing so far :), with the added 
bonus that we don't get updates for all hosts we already know about at 
startup.

Good catch, BTW.