We are in the process of an initial deployment of Solarwinds NPM. I've run into an issue with the SDK and hoping to get some insights as to what might be causing it.
I have a python / rest script that iterates over a number of switch and looks at interfaces that should be monitored. If an interface is being monitored and shouldn't be I delete it. If it is not being monitored and should I add it. The script works as expected except that at some point in the process I get an error from the server that the connection has timed out. At that point the server is unusable for anyone one for a period of time. Sometimes it requires a restart of all of the services to bring it back to a usable state.
I posted the question on Thwack and got a suggestion to do a bulk delete instead of deleting interfaces one by one. I made a copy of the swisclient and added a bulkdelete method. That worked as far as deleting the interfaces but did not solve the problem. I still get the error at some point in the running of the script. The informationservice file indicates that the connection pool may be exhausted. So I modified the script to pause and wait for input between processing each node, thinking I was throwing too many connections too fast. (This will not work in practice, the script needs to run without input but as a diagnostic I thought it would be appropriate step).
I still got the error. When it failed the most recent time it was doing a bulk delete of 108 interfaces from a switch. As a comparison with the GUI I was able to delete 167 interfaces from a node with no problem.
Since the only way I can provoke the error is with an api script, Solarwinds technical support says they cannot help me but did suggest I repost a description of the problem in this forum.
I'm hoping someone can help me figure out how to go about troubleshooting this.
Thanks.