This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.
Is there a problem with Solarwinds monitoring and reporting on Capacity items? My IT Group says yes, and not to trust the reports. My gut tells me they have set things up wrong if there is a problem, and not inherently within Solarwinds.
In order to trust it you have to understand how it gets the info it presents.
In the default settings the node forecast calculations are based on the polled data over the last 7 days. In many cases making a change to either your system resources or workloads over that time would basically invalidate the forecast for that week. You can make it look further back into your node history but it can never be higher than your detailed statistics period, so if you want solarwinds to use a 30 day average then you need to hold detailed stats for 30 days instead of the default 7 (keeping in mind the associated database growth this would cause).
You also need to check if it is reporting based on peak values or average, obviously having a disk hit 100% ever is a problem so you would want to set volumes to forecast against peak daily values. Cpu loads can bounce off 100% periodically and it is generally not a big deal as long as the averages look ok so most people set that to use daily average values for the forecasts.
You can see some documentation about these settings here:
There is also a question of how peaky your workloads are. If your servers only get loaded during business hours and spend all night at 0% load then the default forecasting resource will not reflect your workload accurately, in which case you would have to build yourself a report that takes business hours/days into account.
Here is an example of setting up a report to do that using the old report writer, as of version 12 you can do similar date/time rules in the web based report tool as well.
Another thing that tends to be tricky about disk volumes is that in my experience few disks will ever fill up slowly over time in a predictable way. It tends to be that people get burned when an application runs into a problem and start filling an error log file and the disk goes from 15% to 100% over the course of a few hours. The forecast is not going to be much help in this case and when these do happen they would skew your forecast results for the next week even after you fixed the issue. All you can do there is make sure you have set up alerts to warn when disks have a high percent of space used, and in the past I have set up custom alerts to trigger if a disk starts growing at an unexpected rate, like growing by at least 5% each poll for 3 polls in a row. The nice thing about that particular 3 poll alert logic is that it tends to avoid the issue of admins triggering false alerts when they move a single large file.
Thank you, that was a lot of useful information, and many things I had not thought of.
However I was trying to find out more about monitoring and reporting on Unix/Linus computers.
Our IT group says that Solarwinds does not do a good job on those, and I believe it has to be how they set it up, and not inherent in solarwinds itself.
Specifically with Linux/Unix based systems the failing is usually that lots of appliances and distros have a bare minimal snmpd configuration that doesn't give a lot of the good info people want. I have seen people customize the daemon to give them everything they want but it is not common. If you have SAM you can set up script based monitors that will ssh into the server and parse out basically anything you can dream up, but once again those scripts and the quality of their info tend to vary in quality depending on what specific OS you are running and what commands are available.
Solarwinds has a Linux agent in beta right now so I would expect that to be generally released fairly soon and that may give you more of what you want without having to do a lot of custom work.
