[Linux-ha-dev] Re: LRM Testing

[Andrew Beekhof]

On Jun 3, 2004, at 5:11 PM, Alan Robertson wrote:

> Andrew Beekhof wrote:
>> Hi All,
>> Some issues that resulted from some limited testing today...
>> 1- Location of lrmd executable (should most likely be 
>> /usr/lib/heartbeat like the others)
>> 2- Location of pid file (/var/run is not normally writable by the 
>> haclient user)
>
> Why does it need a pid file?  I wonder...??

probably the same reason the CRM does.... because I copied it from 
somewhere else in the project :)

>
>> 3- lrmd forks and runs in the background, this isnt compatible with 
>> being launched as a "respawn" from ha.cf
>
> Right.  This needs to be fixed... Or at least have an option to enable 
> and disable this behavior...
>
>
>> 4- a bunch of "client_log" messages result from lrmadmin calls.  
>> These should probably be off by default and enabled with -V|--verbose
>> 5- The option --rasupported should probably be renamed 
>> --ra-class-upported and --ra-type-upported <class> be added
>
> What does this do, actually?  I don't see any reason not to just load 
> them all ;-)
>
>> 6- In the admin client generally, the s/type/class/ s/name/type/ need 
>> to be made for consistency
>
> OK.  Where does this show up?  I think we've talked around about this 
> before...

in the code for the admin client and in some of its functions and help 
text

>
>> 7- The LRM is not checking if a RA actually exists during an add.  
>> Since the LRM doesn't tell us the list of all RAs it can support, it 
>> should at least tell us if the RA exists when we do the add.
>
> You should ask for this capability (get a list of all supported).  
> It's easy enough to do, it's just a matter of how to pass it back to 
> you...

i thought this wasnt desirable... since we'd try and cache the info.  i 
thought the idea was to try and start/add the resource and find out 
then if it was allowed.  i'm happy to do either.

>
>> 8- I dont see any version information on the RAs... but I also dont 
>> recall a definite decision on who was responsible for providing it.
>
> OK...  What *kind* of version information?  Shouldn't that be in the 
> metadata?
>
>> 9- After the first failed resource addition, it appears no more can 
>> ever be added until the lrmd is restarted.
>
> Sounds like a bug...  What causes these failures?

commands like below... i suspect 9-11 are related

>
>> 10- It should be legal to have following two resources:
>> lrmadmin -A rsc1 ocf footastic1
>> lrmadmin -A rsc2 lsb footastic1
>> ERROR: on_msg_add_rsc: same id resource exists.
>> 11- The following sequence fails:
>> lrmadmin -A rsc1 ocf footastic1
>> lrmadmin -A rsc2 lsb footastic2
>> ERROR: on_msg_add_rsc: same id resource exists.
>> 12- lrmd[22598]: ERROR: on_recieve_cmd: can not recieve msg
>>     I dont think this is an error, as it appears under normal 
>> circumstances (just before the client disconnect message)
>
> I agree.  It's also misspelled ;-)

and i didnt even notice :)

>
>
> -- 
>     Alan Robertson <alanr at unix.sh>
>
> "Openness is the foundation and preservative of friendship...  Let me 
> claim from you at all times your undisguised opinions." - William 
> Wilberforce
>
> _______________________________________________________
> Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> Home Page: http://linux-ha.org/