2010-07-22 11:43 AM
How are inode full and almost full thresholds determined in Operations Manager 4.0 events?
Is it possible to modify a volume's inode full or almost full threshold? Is there a global option available?
I could not find either in my research.
2010-07-22 08:37 PM
The inodefull and inodenearly full events are generated using the volfull and volnearlyfull threshold values.
They don’t have a threshold value of their own. They use the volfull and volnearlyfull threshold values to generate inodefull and inodenearlyfull events.
Pls add your customer or case to burt 226960
2010-09-29 02:23 AM
I agree with your statement about setting vol/inodes alarm threshold, but I should say I never ever get any alert when the inodes are >80% full (vol warning at 80%). I just got the alarm when the inodes have reached 100%. Is there a way we could monitor the inodes growth in a better way?
2010-09-30 03:08 PM
It's more like first come first sevre. In this example the threshold had breached with respect to the percentage of inodes compared to the percentage of used capacity (Check the screenshot - Full threshold (Hard Limit) set to 22% and Nearly full threshold (soft limit) set to 19 %).
Thanks and regards
2010-09-30 03:57 PM
Thanks for your explanation. If that's the case, may I know what is the magic number here so that we could effectively monitor both volume used capacity and its inodes without throwing our support staff in "crying wolf" situation? At this moment, we had set 90% as nearly full and 95% as full threshold limit, and somehow we did get a notification on inodes breaching over 90%, but DFM didn't send us another alert when it had breached over 95%. We got to know it when it's already inodes 100% utilized.
2010-09-30 04:35 PM
I beileve in your case, the inode usage was so fast, that DFM failed to notice the threshold breach on the "nearly full threshold" value. I believe the it is recognized by the filesystems monitoring interval (15 minutes). With that being said, I would suggest you increase the gap between the nearly full threshold and full threshold values. For example: Nearly full threshold = 75% and Full threshold = 95% (something like that). This should provide you sufficient time to respond back.
Hope that helps.
Thanks and regards
2010-10-01 01:53 AM
I have to contradict you, the inode usage information is gathered by disk free monitor which runs every 30 mins by default and not the fs monitor.
As you said its better to have a considerable difference in the nearly full and full threshold to alert at appropriate time esp in case of indoes as the user might have consumed only 1 inode but could have used by 10% or more of the volume size.