Hi David: I am not exporting data that I am aware of, other than what is required for the Data Collection for Reports. But Reports is not currently a suspect as I have stopped the Data Collection and the issue persists. One thing I should note is that when I was running InterMapper on the Linux VM, I also had IM Flows running. IM Flows is not installed on this server, and my flow rate is typically averaging around 125,000 flows/min. So that traffic volume
I have gone into the Windows Server Resource Monitor and resource utilization is low. CPU is typically less than 10%, Memory is at 39% used, Disk I/O is averaging 200-250Kb/sec, Disk Queue Length is averaging .05, Network is averaging less than 1Mbps. In checking VM Processor in Performance Monitor, CPU Stolen Time is averaging 0.001 In checking the Interface in Performance Monitor, the Output Queue Length, Packet Outbound/Received Discarded/Error counters are zero. In checking the UDP IPV4 in Performance Monitor, the received errors seem to be unrealistically high: Datagrams No Port/sec averages about 17-18 Datagrams Received Errors is averaging about 70,000.000 Datagrams Received/sec is averaging about 44.000 Datagrams Sent/sec is averaging 44.000 Datagrams/sec is averaging about 85.000 In checking the TCP IPV4 in Performance Monitor, Connection Failures also seems unrealistically high Connection Failures is Averaging 240,000.000 Any ideas? Thank you Randy > -----Original Message----- > From: [email protected] [mailto:InterMapper- > [email protected]] On Behalf Of dschnur > Sent: Tuesday, January 31, 2012 6:00 PM > To: [email protected]; intermapper- > [email protected] > Subject: [IM-Talk] Re: Migrated InterMapper from Linux (5.4.5) to Windows > Server 20 > > > Hi Randy, > > > rbaker wrote: > > Any ideas on why we are seeing such a difference in behaviour? > > > As I recall you were exporting to the database; is that still true? If so, > what > does the server's disk I/O load look like? > > When database export is enabled, IM (the server itself, not the database) > writes much more to disk than normal, as it stages data for export. Because > there are significant differences in the way IM interacts with the disk on > Windows versus Linux, it's possible that polling on your new server is > stalling > while waiting on disk I/O. That typically manifests as short-term packet > loss, > since IM isn't counting responses that it perceives as 'too late'. > > We have a special built-in probe used to debug such problems; you can find it > under Experimental -> InterMapper. Create a device with that probe, and let > it run for a while. Then take a look at the status window and please send the > results in an email to support. We'll use the resulting support ticket to > respond to you once we've taken a look at the data. > > A note for others reading this: do not use the InterMapper probe pre- > emptively as a diagnostic tool. It's marked 'Experimental' for a reason, and > is > designed exclusively for debugging specific problems. It doesn't produce > meaningful results on a normal IM server. > > ----------- > David Schnur > Dartware, LLC > > > -------------------- m2f -------------------- > > Read this topic online here: > http://forums.intermapper.com/viewtopic.php?p=3975#3975 > > > > > > __________________________________________________________ > __________ > List archives: > http://www.mail-archive.com/intermapper-talk%40list.dartware.com/ > To unsubscribe: send email to: [email protected] > > > -- > This message was scanned by the Georgian College ESVA and is believed to > be clean. ____________________________________________________________________ List archives: http://www.mail-archive.com/intermapper-talk%40list.dartware.com/ To unsubscribe: send email to: [email protected]
