I have created a dashboard that displays:
panel-1: the disk usage of several mount points of two Linux servers in a simple line chart
panel-2: The prediction date of when disk space may run out if current trending continues as in panel-1.
This works fine when I run it as admin, but when a basic user runs the dashboard, they only see:
- Part of the top of the dashboard page (the Title, the radio buttons, and a caption)
- then "no results found"
Background:
- splunk v6.4.1
- I check a ton of docs, cannot zero in on issue.
- I tried various perm settings nothing works
- I don't know how to include an screenshot image of users results. I tried image?!...path... but no luck
- Could this be b/c I selected "inline search" instead of convert to report?
Perms settings for dashboard are set as:
- Owner= admin
- App = search
- Display for = all apps
- I selected clone to inline search
- I setup perms to be:
Everyone=Read
admin= write
Xml used:
Ray NFS Server DashboardDisk Usage Last 30 Days Last 60 Days Last 90 Days Last 180 Days 30 30 Over last $days$ days ellipsisNone 45 visible visible visible linear linear 0 inherit line 50 10 area gaps minmax 0.01 default shiny all 0 0 ellipsisEnd left Time Terabytes value none 0 1 1 standard absolute after 0 1 true false none 1 0 20 host=mdc4vr1110 OR host=mdc4vr1111 source=df sourcetype=df mount="/*com*" | timechart eval(round((avg(storage_used)/1024/1024/1024/1024),2)) by mount -$days$d@d now Disk Capacity Forecast 0 1 after -1y absolute standard 0 0 [7,14,21,30] ["0xd93f3c","0xf7bc38","0x6db7c6","0x65a637","0x65a637"] 0 none block value search DAYS OF DISK SPACE REMAINING host=mdc4vr1110 OR host=mdc4vr1111 source=df sourcetype=df mount="/*com*" | eval sutbs=(storage_used/1024/1024/1024/1024) | eval sutbsrnd=round(sutbs,2) | eventstats avg(sutbsrnd) AS sutbsavg by mount |fields _time storage_used sutbs sutbsrnd sutbsavg mount | timechart avg(sutbsrnd)| predict avg(sutbsrnd) future_timespan=$days$ algorithm=LLP5 | sort-_time |head 1 | eval predicted_date_epoch=_time | eval timenow_epoch=now() | eval days_remaining_epoch=predicted_date_epoch-timenow_epoch | eval days_remaining=strftime(days_remaining_epoch,"%d") |timechart sum(days_remaining) -$days$d@d now
You are following very bad practices in your search. You should ALWAYS specify index
and sourcetype
. If you fix this, probably it will just work. Your admin
user probably has all indexes
set for indexes searched by default
and your other users have nothing for that value. If this is not it, then you need to give the other users permissions to search against the target index values.
You are following very bad practices in your search. You should ALWAYS specify index
and sourcetype
. If you fix this, probably it will just work. Your admin
user probably has all indexes
set for indexes searched by default
and your other users have nothing for that value. If this is not it, then you need to give the other users permissions to search against the target index values.
Thank you also sir! Yep got that same advice from niketnilay earlier. It did solve my problem.
What are the permissions on the dashboard? Owner, App, or All Apps?
check if the user has access to the indexes involved in the searches
Are you both using the same make/model/version web browser?
Myself and all users use: IE v11.9
@damonmanni add index to your base search query ( like index=abc etc) . Users might not have access to search the required index by default but admin may have that access.
In order to confirm you can ask users to run the following query with days hard coded in the base search ( I have used 30 days):
host=mdc4vr1110 OR host=mdc4vr1111 source=df sourcetype=df mount="/*com*" earliest=-30d@d latest=now
| timechart eval(round((avg(storage_used)/1024/1024/1024/1024),2)) by mount
It is actually a good practice to use index and sourcetype in your base search (even better if you can create a Knowledge Object as eventtype or macro for better maintenance and re-usability)
Excellent suggestion niketnilay. I added index=os to the panel and they now can view it. I had no idea to think about perms/indexes relationship.
Thank you for the quick response.
@damonmanni, I a glad it worked, do consider creating eventtype or macro for your base search with index and sourcetype (also with host and source as per your original search query).