Lars - It is okay and legal to have 2 instances of your plugin specified (as 2 stanzas) in the openhpi.conf file. What distinguishes these stanzas is the unique entity_root value. Your plugin should be designed to handle this. In the case of the first stanza, your plugin would add the stanza's entity_root value to the path of all resources discovered in the simulation file that are associated with this stanza. In the case of the second stanza, your plugin would add that stanza's entity_root value to the path of all resources discovered in the simulation file that are associated with this stanza, and so on. I would attempt to persuade you that the entity_root value does not even belong in the simulation data file. The source of record for the entity_root is openhpi.conf.
Each stanza is unique - in that it has its own entity_root, and its own simulation file. That should allow the plugin to do the right thing - and avoid getting the resources mixed up. In the case of the log files, I'm not sure what to do. Perhaps each stanza should specify its own log file. Does that work? I'm not sure what is the best solution here. Perhaps the plugin should be smart and recoginize a duplicate log file name - and not attempt to open it a second time - but rather just share the file id that was created on the first file open operation. So in this case, a user could set up a common log file that would be shared among the stanzas in openhpi.conf - and the plugin would share the file id for each entity_root that it is using. Or the user could specify unique log file names for each entity_root - and the plugin would use the appropriate one - depending on what resources it is reporting on. I'm just offering some ideas here. Please note that I'm not trying to create more work for you - I'm simply giving you my opinion on how I would expect/want it to work. --michael > -----Original Message----- > From: Lars Wetzel [mailto:[email protected]] > Sent: Thursday, May 13, 2010 2:58 PM > To: Bishop, Michael (ISB Linux/Telco) > Subject: Re: [Openhpi-devel] new simulator - cannot build RPMs > > > Michael, > > you did everything correct. > It is as you wrote and it is an open issue: > In the beginning of the plugin implementation I had it on the > to do list - > Decide if the replacement of the root entry in the simulation > data by the > root entry of the configuration file makes sense. > I forgot it. > So there is still the configuration entry entity path as a > must, but it isn't > used at the moment. The simulator works with the data from the file. > > But now I'm really unsure, if the start of two instances will > work in case the > entity path comes from the configuration file: > I saw also the problem with the logfiles. Both plugin > instances wrote in the > same logfiles. So, if both instances take the first logfile > entry of the > configuration file (both instances have the same plugin name: > new_simulator) > how can I ensure that the second instance will take the > correct entity path > entry? I have to investigate it. > > My opinion: > I think, it could be worth to discuss it separately, if it is > better to take > the root ep from the data file or from the configuration file. > > I will do the change in both direction - remove the > configuration file entry > or replace the data input by the configuration entry. > My favorite it to remove the configuration entry. So you will > be able to run > more than one instance. > > Regards > Lars > > On Thursday, 13. May 2010 22:20, you wrote: > > Lars - > > > > I also tested your new simulator with multiple stanzas in > the openhpi.conf > > file. All I changed from the first stanza to the second > stanza was the > > entity_root value. > > > > On the first stanza, I used: {SYSTEM_CHASSIS, 9} as you show in the > > commented stanza for your plugin. > > > > On the second stanza, I used: {SYSTEM_CHASSIS, 10} > > > > I also added a second simulation.data file call > simulation.data10 - but > > instead of referring to {SYSTEM_CHASSIS, 1} as is the case with the > > simulation.data you included, I changed this to {SYSTEM_CHASSIS, 2}. > > > > While this all seems to work with hpitop and hpitree - > there is something > > weird going on. No where in the output of hpitop or > hpitree is there any > > reference to my entity_root values that I specified in the > openhpi.conf > > file. It seems as though your simulator plugin is ignoring > the entity_root > > values that I am using in the openhpi.conf file. Every > entity path should > > have either {SYSTEM_CHASSIS, 9} or {SYSTEM_CHASSIS, 10} at > the root of the > > path as specified in the openhpi.conf file - but instead, > it appears that > > your plugin is taking the entity_root value directly from the > > simulation.data files. > > > > I'm including my openhpi.conf file, my 2 simulation.data > files, and the > > output of hpitop, and hpitree - so that perhaps you can have a look. > > > > Pehaps I have this misconfigured this. Please let me know. > > > > Regards, > > --michael > > > > Regards, > > Michael Bishop > > Enterprise Servers and Storage (ISB Linux/Telco) > > Hewlett-Packard Company > > 3404 E. Harmony Rd. Bldg. 5L, Post B7, Mailstop 42 > > Fort Collins, CO 80528-9599 > > Phone: 970-898-4393 > > E-Mail: [email protected] > > > > > -----Original Message----- > > > From: Lars Wetzel [mailto:[email protected]] > > > Sent: Thursday, May 13, 2010 1:26 PM > > > To: [email protected] > > > Cc: Bishop, Michael (ISB Linux/Telco); Sutula, Bryan (Open Source > > > Program Office); Andy Cress > > > Subject: Re: [Openhpi-devel] new simulator - cannot build RPMs > > > > > > > > > Hi together, > > > > > > no I didn't test it until 5 min ago: > > > I duplicated the simulation.data file, replaced the EP > root entry and > > > duplicate the libnew_simulator entry in the openhpi.conf file. > > > > > > And it works - with some limitiations: > > > - Both plugin instances use the same logfiles. > > > - I didn't run any test cases > > > > > > A missing feature allows it: The plugin doesn't replace the > > > root - entry of > > > the EP. First I had it on the "to do" list, later I forgot it > > > and now I think > > > it could make sense not to change it. :-) > > > > > > Btw. I will replace the default entry to an absolute path in > > > openhpi.conf.example. > > > I'm not so familiar with configure.in and didn't find an > > > entry to change the > > > openhpi.conf.example, so I will let is at it is. I want to > > > have as less > > > impact as possible on the openhpi configuration with the > new plugin. > > > > > > Regards > > > Lars > > > > > > > > > On Wednesday, 12. May 2010 22:31, Bishop, Michael (ISB > > > > > > Linux/Telco) wrote: > > > > > This seems reasonable if the simulator behaves like the other > > > > > plug-ins. > > > > > Does this mean that you can run several instances of the > > > > > > new simulator > > > > > > > > with different data files for each, assuming you have > > > > > > multiple stanzas > > > > > > > > in the openhpi.conf file? > > > > > > > > > > Bryan > > > > > > > > Good question, Bryan. > > > > > > > > Lars have you tested your new simulator with multiple > stanzas in the > > > > openhpi.conf file - with each stanza specifying a different > > > > > > simulation.data > > > > > > > file? Seems like this should work - if your design follows > > > > > > the normal > > > > > > > rules for plugins. > > > > > > > > --michael > > > > > > > > > -------------------------------------------------------------- > > > > > ---------------- > > > > > > > > > > _______________________________________________ > > > > > Openhpi-devel mailing list > > > > > [email protected] > > > > > https://lists.sourceforge.net/lists/listinfo/openhpi-devel > > > > > > -------------------------------------------------------------- > > > ------------- > > > > > > >--- > > > > > > > > _______________________________________________ > > > > Openhpi-devel mailing list > > > > [email protected] > > > > https://lists.sourceforge.net/lists/listinfo/openhpi-devel > > > > > > -- > > > ------------------------------- > > > Dipl. Wi.ing. > > > Lars Wetzel > > > Uttinger Str. 13 > > > 86938 Schondorf a. Ammersee > > > > > > Tel.: 0179-2096845 > > > Mail: [email protected] > > > > > > USt-IdNr.: DE181396006 > > -- > ------------------------------- > Dipl. Wi.ing. > Lars Wetzel > Uttinger Str. 13 > 86938 Schondorf a. Ammersee > > Tel.: 0179-2096845 > Mail: [email protected] > > USt-IdNr.: DE181396006 > ------------------------------------------------------------------------------ _______________________________________________ Openhpi-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/openhpi-devel
