Dependency Issues

The contract my company is on manages thousands of devices. We recently had a power outage that caused a large chunk of buildings on one of our campuses to go down, as a result of these buildings and devices going down at once made our Solarwinds site crash. To prevent this kind of thing happening in the future I decided to build dependencies for the buildings on that campus. About a month later we had another power outage and to my surprise the dependencies didn't always work. Some worked properly and I could see the groups in the "Unreachable" state, but other buildings were in the "down" state that I was trying to avoid; others were in the warning state as well... From what I understood, if the dependencies are build from the building switches all down streamed devices should change to "Unreachable" and this was not the case... Anyone have any ideas? Any help is appreciated. Thanks

Find more posts tagged with

orion

dependencies

solarwinds

dependency

Accepted answers

Jfrazier

Ah...but likely a timing issue.

If it had just polled the parent device and a minute later the power went out it could be polling the child devices before it gets back to the parent.

ICMP and SNMP don't report on the path they take. They either return or not and polling is updated from that.

All comments

Jfrazier

Are you using groups in your dependencies ?

FormerMember

Yes, I was setting the building switch as the parent and having all down stream devices in groups based on what building they were in.

Jfrazier

Okay so you should have gotten a down event for the switch unless it was on UPS.

Are there other alerts that encompass your building/devices that could have fired as well?

FormerMember

Yeah, I was expecting the dependency to work like this: The parent switch shows "down", all devices in the group (the child) show "unreachable" instead of down. I have alerts set up to go off on downed devices, but not for unreachable. This is not how some of the dependencies worked. Some worked properly and others the parent and all child devices showed "down".