Basically it says that average queue size is 234, average execution time is 152 ms and at the moment it runs about 1200 IOPS at the moment. Is it normal load? I'm not responsible for the hardware but I guess we have RAID 5.
When I start netflow on one device for one interface, number of files in shared folder starts to grow consistently. Also IQ_SYSTEM_TEMP dataspace starts to grow quickly.
I talked to out technicians and they've asked me what block size SYBASE operate? I can't find this parameter anywhere. Any suggestions?
Also I would like to know where is log file for SYBASE database. I can't find exact reason why it crashes.
Any tips on troubleshooting such behavior would be appreciated.
While I don't have a sybase related answer, I can however imagine a scenario where this could take place.
>>"Is it normal load? I'm not responsible for the hardware but I guess we have RAID 5.
When I start netflow on one device for one interface, number of files in shared folder starts to grow consistently. Also IQ_SYSTEM_TEMP dataspace starts to grow quickly."
I am making a few assumptions here, and please correct me if I am wrong, but when you say you start iSPI Traffic, you mean to say that you start the Traffic Master and Traffic Leaf services, right?
The Traffic Leaf has a setting, for each collector, to store received flows in files.
If Traffic Leaf is on the same server as iSPI Metrics, the everytime Leaf receives a netflow packet, it will store the info (open any leaf collector, check the option that says "store flow in file", this must be set to false unless you have it set to true for some diagnostic purpose).
Second, iSPI Traffic Master has a setting which determines the rate at which Master writes data to the NNMi Shared Folder (%NNMDataDir%shared/perfSpi/datafiles/metric/final and a couple other sub dirs).
The setting is governed by time (2-3 minutes I think), and by a number (100000 by default).
Essentially, once Master has processed 100000 records, or 2 minutes have elapsed, (whichever happens earlier) Master will flush data (in the form of a CSV GZ file) to the NNMi shared folder.
The setting for master is configurable on the Master UI at Master Settings.
If the setting is too low (100 for example), then Master will frequently write data to NNMi Shared dir, which will increase the IO load on NNMi, and since ETL (Extract-Transform-Load) processes from iSPI Metrics have to pick up this data, they will also be stressed.
Just a couple of theories, in the mean time, let me see if I can give you some more info on SYBASE block size.
>>"Also I would like to know where is log file for SYBASE database. I can't find exact reason why it crashes. "
The prspi.log would be best best, if iSPI Metrics Sybase (ASE/IQ) is crashing (%NPSDataDir%NNMPerformanceSPI\logs\). Search for strings such as FATAL (show stoppers in most cases), or SEVERE, or WARN*(WARN/WARNING) etc.
A little light reading from this link may also help (now or in future):