#16 Graph shows dips in all collected metrics

[Johan Nilsson]

Using asemon_logger 3.0.6 on as ASE15.7, all graphs on Summary-page is showing dips every 5 or so minutes.
The problem occurred when switching primary server in a MSA-setup, the old primary didn't have this behaviour.

Have looked at the $server_Engines-table, and there's an entry for each engine, for every sample, but they differ significantly from the surrounding values, i.e.

1> select Ts=convert(varchar,Timestamp,109), avgCpu=avg(convert(float,UserCPUTime1000)/Interval100) from $server_Engines where timestamp > dateadd(hh,-1,getdate()) and contextswitches > 0 group by timestamp
2> go
Ts avgCpu

---

Nov 10 2017 10:52:22:463AM 56.170425
Nov 10 2017 10:53:22:520AM 62.439642
Nov 10 2017 10:54:22:606AM 3.952602
Nov 10 2017 10:55:23:023AM 69.520310
Nov 10 2017 10:56:24:253AM 60.425921

1> select Ts=convert(varchar,Timestamp,109),UserCPUTime from SD2_Engines where timestamp > dateadd(mi,-15,getdate())
2> go
Ts UserCPUTime

---

Nov 10 2017 10:53:22:520AM 36
Nov 10 2017 10:53:22:520AM 39
Nov 10 2017 10:53:22:520AM 38
Nov 10 2017 10:53:22:520AM 38
Nov 10 2017 10:53:22:520AM 38
Nov 10 2017 10:53:22:520AM 36
Nov 10 2017 10:53:22:520AM 39
Nov 10 2017 10:53:22:520AM 36
Nov 10 2017 10:54:22:606AM 2
Nov 10 2017 10:54:22:606AM 3
Nov 10 2017 10:54:22:606AM 2
Nov 10 2017 10:54:22:606AM 2
Nov 10 2017 10:54:22:606AM 2
Nov 10 2017 10:54:22:606AM 3
Nov 10 2017 10:54:22:606AM 2
Nov 10 2017 10:54:22:606AM 3
Nov 10 2017 10:55:23:023AM 44
Nov 10 2017 10:55:23:023AM 42
Nov 10 2017 10:55:23:023AM 41
Nov 10 2017 10:55:23:023AM 44
Nov 10 2017 10:55:23:023AM 40
Nov 10 2017 10:55:23:023AM 41
Nov 10 2017 10:55:23:023AM 43
Nov 10 2017 10:55:23:023AM 41

what could be the issue? We have restarted the asemon-logger several times, without any errors in the log, and without any improvements.
Any help is greatly appreciated.

Thanks in advance
Johan Nilsson

[Jean-Paul Martin]

Hi Johan, sorry to answer soo late
dips in graphs means there were missing collects for these timestamps

Missing collects can be due to :

• monitored ASE is overloaded and asemon queries are delayed
• asemon_logger has not enougth resources to execute (CPU or memory) and misses some collects
• ASE managing the repository database (archive database) is overloaded and asemon_logger is delayed when inserting data
• asemon_logger is stopped

Anyway, now, since asemon_report V3, you can see holes is collects. Previous asemon_report didin't show these holes
Asemon_report consider a "hole" when 2 collects have a timstamp difference of 2 X collect's delay

Best regards
Jpm

[Johan Nilsson]

Hi Jean-Paul,

Thaks for your response.
Well, the values aren't missing, every 5 minutes or so, they just differs dramatically from the surrounding values, as seen in the output from the select on the reporting ASE.

We have these dips after switching primary server, we didn't see this behaviour before switching primary ASE-server. And the load on the monitored server is "manageble", peaking on approx 80% cpu-load.

We have a switch of primary server scheduled for monday, I'll update the issue after the switch.

Best regards
Johan