Warning or Critical status is not cleared when the situation normalizes after installing PSP 8.0

[Andrea Giangone]

Since installing PSP 8.0 all servers are generating an excessive amount of alerts. It seems that this issue once happened with PSP 7.4 but it may have now reoccurred.
Specifically, it does not look like the SNMP 15 time cycle is respected but rather that alarms are sent as soon as the threshold is reached. This is valid for Processor as well as Logical disk Usage.
The second issue is that the critical state in SNH does not clear when the usage falls back below the threshold.

[KarloChacon]

hi Andrea

initially which PSP did servers have before 8.0? PSP 7.91? PSP 7.9?

for me PSP 8.0 has quite weird behaviors... most of the time when doing big jumps between PSPs example 7.6 - 7.7 -> 8.0

regards

[Andrea Giangone]

Hi Karlo,

all three servers I ventured out installing PSP 8.0 had 7.91 before. It is clear that the issues with the high number of alarms stems from the fact that there is no time wait applied. BTW my default SNMP poll time is set for 2 minutes so a threshold should be exceeded for at least 30 minutes before triggering an alarm.

[Andrea Giangone]

Please see the reply received from Joel Rubenstain at HP:
"Hi Andrea,

I have been doing some testing with the 7.91 and 8.0 agents.

I have observed that the 7.91 agents do not send a trap for Processor Time or LogicalDisk Busy Time thresholds being exceeded until 15 data collection intervals (default = 2 minutes * 15 = 30 minutes) have elapsed and that the 8.0 agents send the trap when the threshold has been exceed for 1 data collection interval. I will be checking with our engineering folks to determine if this was a deliberate design change.

However I have observed that when the Processor Time or LogicalDisk Busy Time drops below the set thresholds the overall server status returns to normal. I have also observed that only 1 trap is sent as long as the thresholds are being exceeded over several data collection intervals and another trap will not be sent until the utilization drops below the threshold and then exceeds it again. Is this the same behavior you are seeing?

[Andrea Giangone]

Yes that is the same behavior.

[KarloChacon]

oh

it seems like your issue was escalated to higher level uh?

regards

[Andrea Giangone]

Well, this is a real issue because we continuously miss important alarm due to the flood. I'm glad to see that HP was responsive and at least acknowledge my initial suspicion. Now I'm hoping to see a fix in a few days.