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

Ask em to take your 96 checks and make them a stored proc that COUNTS the problems, then lists a summary Run the stored proc in a SQL UX monitor, return count of problems and summary, set threshold on count But like nah not really

Do you have a block of code after that to write-host message and statistic? If no, add that first if yes, probably worth adding that code block. In general it doesnt tend to like linebreaks, or arrays of unknown size, you can deal with it by using a string character and then REPLACE()ing that for a linebreak on display,…

I think you'd go hardware item to host/node to host to datacenter to vcenter or similar

There's no health equivalent off SWO, and some of those mini charts you couldnt do, but most of the columns are doable otherwise. The tabs at the top maybe are in the current RC release, not clear to me yet

If you want to actually count stuff, it's COUNT(columnname) .... GROUP BY Something if need be

msseversum's got some resources over here thwack.solarwinds.com/.../custom-sql-swql-for-alerts-history---sla-reporting

Reports -> Create report -> Browse for Logo?

Settings -> Web console settings -> Logos on there I think, references the filepath so you can probably pull a backup from that location

That join aint right I've got stuff like this in prod, but the query can look quite a bit different based on environmental specifics I think you want to join to at least a couple other alerts tables, and then to ContainerMembers where the memberID = AlertObjectID or similar. Note if you've got groups in groups this can get…

The answer to question 2 is "Yes"

Feels like you've set something up funny. Not sure which thing. EG, your alert says not up. But you've kept the defaults for CPU/Mem thresholds, which will translate the status into a warning/critical, which will trigger your alert Alternately, the test goes unknown because creds are negotiated poorly or something, grey…

Best short term fix: Check in mib browser to see if you can find the data you're expecting If you can, set a UNDP for that, assign to the palos, then report to SLW If you cant, report to palo alto.

Will have a play with this, wont be super soon, thanks for sharing

That is a level of broken i'm glad to have not experienced! Reboot/reset everything/ check patches or something/check security software/service accounts/event logs/stuff like that Call support (would have assumed firewall too, but the fact that dies when in use and comes back when not in use doesnt sound firewalley)

You want to have an eye like this for the equivalent on all the other layers that are important to you

Couple things i'd recommend changing here Use the scope section to define your application name, not the trigger, it'll mess up stuff like the "all alerts this could trigger" widgets later down the line The two application statuses can be in the single OR block, no need for two (bonus one - Usually Unknown and Down end up…

As much as I can and I want to 1000x my use of it this year, but in the last month or two not quite daily, which is weird. We're approaching a world where you can ask for text and get it (and images, video, depth maps, etc). Commands to systems and configs and everything is text. No end to the improvements that'll get…

Note, doing one alert per component isnt ideal, unless that specific component really needs something custom

Referencing your snip above, You want "AND Status is equal to down" in the trigger condition, and ID (or whatever else) equals the ID of the thing. You can get that from the URL on the page you're on Or like Name = X and Server name = Y in the Scope section above.

Ok that makes more sense, was looking at "it still returns the data as bps and not percents" To not get a generic error, pop it in swql studio SELECT I.InterfaceID, I.InterfaceSpeed, N.NodeID, N.Caption AS NodeName, I.Caption AS Interface_Caption, AverageOutboundBps90th / I.InterfaceSpeed AS PercentUtilizationOutbound90th…

I dont get how that doesnt output 90, but if it's a string/interger type problem, multiply the string by 1

Not sure what your ideal data would look like, could you mock up an example or elaborate? (If the 90th percentile was selected as a percentage, it'd show 90 i guess?)

You're in the wrong forum channel, you want the WPM product forum not the WHD one In the WPM forum you'll find some posts from me on this topic right at the top. If you have low security it's really easy, if you have moderate to high security it's very hard without additional supporting kit.

In addition to what KMSigma said, you can pretty much just use formula as you mentioned as long as if you're SELECTing it you give it a column name (AS x). If you were filtering with it you can probably just whack something like that in the WHERE. Sometimes you may need a subquery but doesnt seem like you'd need anything…

Oh no worries Steve I was replying to the OP too!

Start from ContainerMembers rather than Nodes then and you'll probably arrive at the right answer 

Group name doesnt exist like that. You're thinking about "where in this group" but what you're asking for is specifically the property against each device to match 'PI Servers'. Think what it'd be like if a server was a member of two groups. What you want to do is SELECT or JOIN to the Orion.ContainersMembers where…

See above^

This wont hit a remote server. Also wont do SQL But the SQL equivalent is in Classic dashboards as Custom Query or something like that

I think what you're looking for is the SAM SQL user experience monitor, but also that it wont be as straightforward as you'd like