Description of problem: I've set this configuration: Group "Internal Default Group": Group ID: 1 Name: Internal Default Group Features (priority: name): 0: Master 1: NodeAccess 2: ExecuteNode 3: HAScheduler 4: HACentralManager Parameters: HA_LOCK_URL = file:/mnt/qa/scratch/MRG/testlocal CONFIGD_OVERRIDE_DIR = /root/condor REPLICATION_LIST = _ips:ports_ .. ALLOW_READ = * HAD_LIST = _ips:ports_ ... SPOOL = _directiry_ SCHEDD_NAME = ha-schedd@ ALLOW_WRITE = * CONDOR_HOST = test 1) Feature ExecuteNode requires non-empty value of CONDOR_HOST defined by user. But HACentralManager should erase CONDOR_HOST because it has to be not defined for this feature. HACentralManager doesn't erase CONDOR_HOST. In this case I've solved it by overriding CONDOR_HOST by define it in configuration in CONFIGD_OVERRIDE_DIR. 2) COLLECTOR_HOST is required for HACentralManager and has to be set to list of potential central managers. But now it is defined as: COLLECTOR_HOST = $(CONDOR_HOST) But CONDOR_HOST has to be undefined so COLLECTOR_HOST is undefined/empty too. I think it has to be defined by user during set of HACentralManager feature. Version-Release number of selected component (if applicable): ruby-wallaby-0.10.4-2.el5 wallaby-utils-0.10.4-2.el5 python-condorutils-1.4-6.el5 python-wallabyclient-3.9-3.el5 condor-wallaby-client-3.9-3.el5 condor-wallaby-tools-3.9-3.el5 condor-wallaby-base-db-1.5-2.el5 condor-7.4.5-0.8.el5 wallaby-0.10.4-2.el5 How reproducible: 100% Steps to Reproduce: 1. set up remote configuration 2. set up these features: Master,NodeAccess,ExecuteNode 3. set up this feature: HACentralManager Actual results: It is not possible to set up ExecuteNode and HACentralManager on same node. Expected results: It will be possible to set up ExecuteNode and HACentralManager on same node. Additional info:
1) CONDOR_HOST is just a macro used to set other configuration params, usually security based settings. It doesn't actually "need" to be anything, and we use it effectively interchangeably with COLLECTOR_HOST. In the default db, CONDOR_HOST is set to "" in the HACentralManager feature, and this should be a 0 to force the user to set the value. However, if CONDOR_HOST is set by a higher priority feature, as in this case (ExecuteNode), then it doesn't matter what HACentralManager sets it to as the node's configuration will either get the value from the higher priority feature, or will need to set the value for CONDOR_HOST explicitly. The behaviour witnessed does not appear to be incorrect. However, HAcentralManager will be changed such that CONDOR_HOST is required to be set by the user. A better long term solution is likely to get rid of CONDOR_HOST all together. 2) The value of CONDOR_HOST, and thus COLLECTOR_HOST, will be dependent upon what other features are set on the node. Again, setting CONDOR_HOST to be input by the user will allow the HACentralManager feature to work alone on a node, but does not remove the need to be careful with priorities and param values when HACentralManager is set on a node/group with any other feature that sets CONDOR_HOST. Fixed on master.
Technical note added. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. New Contents: C: The default value for CONDOR_HOST is incorrect in the default database for the HACentralManager feature. C: The user will need to explicitly set CONDOR_HOST when enabling HACentralManager on a node without the tools asking for a value. F: The value for CONDOR_HOST will be prompted for by the tools when setting HACentralManager. R: HACentralManager will ask to set CONDOR_HOST when enabled on a node/group.
Technical note updated. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. Diffed Contents: @@ -1,4 +1 @@ -C: The default value for CONDOR_HOST is incorrect in the default database for the HACentralManager feature. +Previously, the default value for CONDOR_HOST was set incorrectly in the default database for the HACentralManager feature. Due to this issue, users had to explicitly set CONDOR_HOST when they enabled HACentralManager on a node without the tools asking for a value. With this update, the tools prompt for the value for CONDOR_HOST when HACentralManager is set. Now, HACentralManager asks to set CONDOR_HOST when enabled on a node/group.-C: The user will need to explicitly set CONDOR_HOST when enabling HACentralManager on a node without the tools asking for a value. -F: The value for CONDOR_HOST will be prompted for by the tools when setting HACentralManager. -R: HACentralManager will ask to set CONDOR_HOST when enabled on a node/group.
Technical note updated. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. Diffed Contents: @@ -1 +1 @@ -Previously, the default value for CONDOR_HOST was set incorrectly in the default database for the HACentralManager feature. Due to this issue, users had to explicitly set CONDOR_HOST when they enabled HACentralManager on a node without the tools asking for a value. With this update, the tools prompt for the value for CONDOR_HOST when HACentralManager is set. Now, HACentralManager asks to set CONDOR_HOST when enabled on a node/group.+Previously, the default value for CONDOR_HOST was set incorrectly in the default database for the HACentralManager feature. Due to this issue, users had to explicitly set CONDOR_HOST when they enabled the HACentralManager on a node without the tools asking for a value. With this update, the tools prompt for the value for CONDOR_HOST when the HACentralManager is set. Now, the HACentralManager asks to set CONDOR_HOST when enabled on a node/group.
Tested with: condor-wallaby-base-db-1.12-1 condor-wallaby-client-4.0-6 wallaby-utils-0.10.5-4 wallaby-0.10.5-4 ruby-wallaby-0.10.5-4 condor-wallaby-tools-4.0-6 python-wallabyclient-4.0-6 Tested on: RHEL5 i386,x86_64 - passed RHEL6 i386,x86_64 - passed CONDOR_HOST was set to list of all central managers >>> VERIFIED
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on therefore solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHEA-2011-0889.html