Discussion:
[Pacemaker] OCF_RESKEY_CRM_meta* envars
Florian Haas
2009-05-28 20:32:17 UTC
Permalink
Andrew,

would you mind pointing me to wherever in the code the
OCF_RESKEY_CRM_meta* variables are set and passed into the RA
environment? I'd like to understand where and how this happens, have
been unable to find documentation, and a simple recursive grep in the hg
checkout wasn't too enlightening either.

Thanks!
Cheers,
Florian
Andrew Beekhof
2009-05-29 07:52:44 UTC
Permalink
Post by Florian Haas
Andrew,
would you mind pointing me to wherever in the code the
OCF_RESKEY_CRM_meta* variables are set and passed into the RA
environment? I'd like to understand where and how this happens, have
been unable to find documentation, and a simple recursive grep in the hg
checkout wasn't too enlightening either.
Check out the action2xml() function, in particular its use of hash2metafield().
Florian Haas
2009-06-02 08:24:30 UTC
Permalink
Andrew,
Post by Andrew Beekhof
Post by Florian Haas
Andrew,
would you mind pointing me to wherever in the code the
OCF_RESKEY_CRM_meta* variables are set and passed into the RA
environment? I'd like to understand where and how this happens, have
been unable to find documentation, and a simple recursive grep in the hg
checkout wasn't too enlightening either.
Check out the action2xml() function, in particular its use of hash2metafield().
Hm. I feel a wee bit smarter after looking through that, but still fail
to understand why these environment variables seem to be available to
the RA only during some operations (notify?), while unavailable in
others (monitor?). Forgive my ignorance. Can you shed a little more
light on that?

Thanks!
Cheers,
Florian
Andrew Beekhof
2009-06-02 09:15:02 UTC
Permalink
Post by Florian Haas
Andrew,
Post by Andrew Beekhof
Post by Florian Haas
Andrew,
would you mind pointing me to wherever in the code the
OCF_RESKEY_CRM_meta* variables are set and passed into the RA
environment? I'd like to understand where and how this happens, have
been unable to find documentation, and a simple recursive grep in the hg
checkout wasn't too enlightening either.
Check out the action2xml() function, in particular its use of hash2metafield().
Hm. I feel a wee bit smarter after looking through that, but still fail
to understand why
Do you mean technically or conceptually?
Which vars in particular are you talking about?
Post by Florian Haas
these environment variables seem to be available to
the RA only during some operations (notify?), while unavailable in
others (monitor?). Forgive my ignorance. Can you shed a little more
light on that?
Thanks!
Cheers,
Florian
_______________________________________________
Pacemaker mailing list
Pacemaker at oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Florian Haas
2009-06-02 09:24:17 UTC
Permalink
Andrew,
Post by Andrew Beekhof
Post by Florian Haas
Post by Andrew Beekhof
Check out the action2xml() function, in particular its use of hash2metafield().
Hm. I feel a wee bit smarter after looking through that, but still fail
to understand why
Do you mean technically or conceptually?
I would say I need to understand conceptually first, before I can tackle
this issue on technical terms. :)
Post by Andrew Beekhof
Which vars in particular are you talking about?
OCF_RESKEY_CRM_meta_clone_max
OCF_RESKEY_CRM_meta_clone_node_max
OCF_RESKEY_CRM_meta_master_max
OCF_RESKEY_CRM_meta_master_node_max

I fail to understand under which circumstances exactly these are passed
in for a Master/Slave resource. It would make sense to me that the would
be passed in on any operation, but it appears to me that they are in
fact not available in some.

Your insight is much appreciated. Thanks!

Cheers,
Florian
Andrew Beekhof
2009-06-02 11:48:22 UTC
Permalink
Post by Florian Haas
Andrew,
Post by Andrew Beekhof
Post by Florian Haas
Post by Andrew Beekhof
Check out the action2xml() function, in particular its use of hash2metafield().
Hm. I feel a wee bit smarter after looking through that, but still fail
to understand why
Do you mean technically or conceptually?
I would say I need to understand conceptually first, before I can tackle
this issue on technical terms. :)
Post by Andrew Beekhof
Which vars in particular are you talking about?
OCF_RESKEY_CRM_meta_clone_max
OCF_RESKEY_CRM_meta_clone_node_max
OCF_RESKEY_CRM_meta_master_max
OCF_RESKEY_CRM_meta_master_node_max
I fail to understand under which circumstances exactly these are passed
in for a Master/Slave resource. It would make sense to me that the would
be passed in on any operation, but it appears to me that they are in
fact not available in some.
You appear to have found a bug.
Those should always be present.
Post by Florian Haas
Your insight is much appreciated. Thanks!
Cheers,
Florian
_______________________________________________
Pacemaker mailing list
Pacemaker at oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Florian Haas
2009-06-02 12:04:04 UTC
Permalink
Post by Andrew Beekhof
Post by Florian Haas
Post by Andrew Beekhof
Which vars in particular are you talking about?
OCF_RESKEY_CRM_meta_clone_max
OCF_RESKEY_CRM_meta_clone_node_max
OCF_RESKEY_CRM_meta_master_max
OCF_RESKEY_CRM_meta_master_node_max
I fail to understand under which circumstances exactly these are passed
in for a Master/Slave resource. It would make sense to me that the would
be passed in on any operation, but it appears to me that they are in
fact not available in some.
You appear to have found a bug.
Those should always be present.
OK. In that case let me see how I can reliably reproduce this problem.

Cheers,
Florian

Loading...