adam.beedell · SRE/Machine Learning Student · ✭✭✭✭✭

GPT's method worked Usually you can get away with ~: Select x.parent.name, Count(ID) as [Count], From X Group by x.parent.name

Got some choices for datasource, just use the first one that works well NPM to the host - Possible sometimes not others NPM to the idrac/ilo/whatever - Usually works VMAN Hardware sensors through parent vcenter/hyperv/if standalone - Pretty normal VMAN Alarms reading parent vcenter - Can be annoying VMAN Status reading…

Should be able to get some stuff out via NPM alerting. I've been starting work in this space recently for NCM and SCM. Anything that'd be helpful to share if I end up working on it?

This is a long road to walk. You'll realise that the issue is the definition of "down" kinda exists in an audience's minds whereas you're working with some software Some stuff i'd recommend: Set up a process in which outages are tracked as per major incident processes. Uptime = 100% - Sum of incident duration. Have access…

SELECT TOP 100 NodeID, HostName, SystemName, Domain, DNSName, DomainRole, AssetType, OperatingSystem, OSVersion, OSArchitecture, ServicePack, OSLanguage, LastBoot, HardwareSerialNumber, ProductNumber, Manufacturer, Model, LastLoggedInUser, WarrantyDate, CountryCode, LastWarrantyPoll, DeviceTimeZone, AntivirusStatus,…

It's in Asset Inventory if polled, eg: Windows 2022 Server10.0 (Build 20348 Multiprocessor Free)

Also, you could pull stuff out of the DB if you really had to

Not sure, I've got the export button too, suspect the same as you, need to get a lab back up. I still had an export button with a low-rank account so it's not from Admin or anything. I expect it comes down to if you've got the proper log product or the free one. Would confirm with support. My guess is @"emalto" you've got…

If you're lookin in sql, look in swql

Creds probably

Should work, either something is broke, misconfigured, or you're not allowing for enough time for a stats/topology collection

You'd want to do this directly in the script, but yeah totally doable. Start with one of the templates files and ask chatgpt for an amendment, probably faster than finding a perfect example

Is the 2nd example exactly what the 1st example gives back?

(We've all got day jobs, cant be spending all day writing SWQL for people all the time I'm afraid, but I'll paste stuff in when I've got it to hand or if diving into it is interesting enough)

1st go making a custom alert for this, and with baselines in mind rather than pure change A macro like this produces a correct URL, however it's making some unsafe assumptions at the moment - Alerting at the node level but having potentially several profiles and elements that could change means there's a many-to-one…

I'll probably be giving this a go in the next few days, I dont expect it's straightforward. Pretty sure you can generate the link to this automatically though. Yell if you've got anywhere with it since you posted, didnt see it at the time

https://github.com/solarwinds/OrionSDK/wiki/SWQL-Functions Available SWQL datetime functions here, the actual DATETIME one isnt well documented but we can add to it Is the polled data in non-US date format? You get an error passing in too many months/half the non-US-format dates

Make sure the SQL DB recommended settings are the same, particularly the collation Backup -> restore -> config wiz

Looks like you can Join Orion.APM.ComponentTypeMappingStrategy ctms on x.componenttpe = ctms.componenttype Otherwise case statement I suppose

Short answer: "Yes" Medium answer: Kinda depends on all sorts of stuff, but ultimately if you've got that error there's something on the request sent to the device that the device said no to. I'd try creds first.

Something like ${N=SwisEntity;M=ComponentAlert.ComponentMessage} as @"weston.hood" said is the way to go If you had an issue with that one, then you could use SWQL, then if you cant use that one, then use SQL. SQL should be waaaay down the list as it may break during patching

I dont know SWO alerting well, but you should consider working on discards as percent of traffic speed, some discards are fine, and you'll have loads of noise for high-use links with low static thresholds

Yaqua's link is better, but- 1) $credential = ${CREDENTIAL}, then you can extract the user/pass if you need to 2) You kinda can if you really want to, cURL's probably installed - @ UX guys, a cURL component would save a lot of time. 3) Yeah that's what I do, -h is headers and you do this: $headers = @{ Authorization =…

Feels certificate-y or security-ey. Probably someone's turning ya off.

You probably want to cut that list down a bit rather than just all. Draws false connections. An example similar macro is here: Other downs in the last 15 minutes ${SQL: SELECT Caption+Char(9)+Char(10) FROM Nodes WHERE Vendor in ('Windows', 'Linux', 'net-snmp') AND Status like '2' AND ISServer like '1' AND DATEDIFF(minute,…

Yeah there's a formatter for that ~

Can do day() >1 and day() <6 too for usual weekends

Roughly: Where datetime > x and datetime < y and hour(datetime) > 9 and hour(datetime) < 18

As a percent or bps or something yeah sure

Ah, I think I mistested that the first time "A similar query on Orion.Nodes returns data fine using 'username@domain' format." "the query works fine when using DOMAIN\username format to log in." tested both this time - results with domain\username format, no results with username@domain format, likely bug