Node scan does not finish with failing async service detectors
Description
In my OpenNMS installation I provision nodes via OpenNMS provisioning and was using the default service detectors. However, after upgrading to OpenNMS 1.12.7 from 1.12.1 I observed the problem that only the explicitly provisioned IP interface of a newly provisioned router node was provisioned but the other interface were not automatically added as before.
From the provisiond log files it seems clear that the node scan never finishes and to me looks like the reason is that the async service detectors (HTTP, HTTPS,...) never complete their second attempt. E.g. the last log statement for HTTPS in provisiond.log is:
The log statement for the failure of the second attempt to detect the service is missing. E.g from a 1.12.1 installation:
I have now changed the service detectors to only the services which are present on the nodes and the node scan is then successfully completed (In general I had planned to keep configuration changes to a minimum, therefore I had not changed the service detector configuration before.).
I can reproduce this issue with a fresh 1.12.8 installation in a VM with JDK 1.7.0_45 and PostgreSQL 9.1.0 by creating one provisiong group with one node with a not existing IP address and performing a synchronize.
There's not enough info here to reproduce, closing issue.
Seth Leger June 28, 2016 at 10:24 PM
Hi Christian,
Please add any more feedback that you have to this issue. Otherwise, I'll have to close it as cannot reproduce. Thank you for your input.
Christian Huter April 9, 2015 at 1:50 PM
Thank you for reporting back on this issue! We have not yet updated to 15.0 (we are using 14.0.3). As soon we have updated, I will try to reproduce this issue with 15.0.x. But I think I will not be able to do the update before end of May.
Seth Leger April 9, 2015 at 11:48 AM
Can you reproduce this with version 15.0.1 (or 15.0.2, it should be released shortly)? We thought that the issue was resolved in that version of 1.12.
In my OpenNMS installation I provision nodes via OpenNMS provisioning and was using the default service detectors. However, after upgrading to OpenNMS 1.12.7 from 1.12.1 I observed the problem that only the explicitly provisioned IP interface of a newly provisioned router node was provisioned but the other interface were not automatically added as before.
From the provisiond log files it seems clear that the node scan never finishes and to me looks like the reason is that the async service detectors (HTTP, HTTPS,...) never complete their second attempt. E.g. the last log statement for HTTPS in provisiond.log is:
The log statement for the failure of the second attempt to detect the service is missing. E.g from a 1.12.1 installation:
I have now changed the service detectors to only the services which are present on the nodes and the node scan is then successfully completed (In general I had planned to keep configuration changes to a minimum, therefore I had not changed the service detector configuration before.).
I can reproduce this issue with a fresh 1.12.8 installation in a VM with JDK 1.7.0_45 and PostgreSQL 9.1.0 by creating one provisiong group with one node with a not existing IP address and performing a synchronize.
This issue may be related to http://issues.opennms.org/browse/NMS-6412 which has been marked as fixed in 1.12.6.