sflow 1 destination 10.0.0.50 owner 1 timeout 4294967295
sflow 1 polling ethernet 1/g1-1/g10 30
sflow 1 sampling ethernet 1/g1-1/g10 512
A previous posting discussed the selection of sampling rates. Additional information can be found on the Dell web site.
See Trying out sFlow for suggestions on getting started with sFlow monitoring and reporting.
This syntax didn't quite work for me - I'm using firmware 3.1.5.2 on the M6220 powerconnect switches. These are blade switches for the M1000E Chassis.
ReplyDeleteWhat worked for me was
sflow 1 destination
sflow 1 destination owner timeout 4294967295
sflow 1 polling ethernet 1/g1-1/g16 30
sflow 1 polling ethernet 2/g1-2/g16 30
sflow 1 sampling ethernet 1/g1-1/g16 1024
sflow 1 sampling ethernet 2/g1-2/g16 1024
Notice that the sflow destination ip and the owner lines needed to be separate. Also, with the version I'm running, 1024 was the smallest sample size it would allow.
Another btw, if you run a stack on two powerconnects, the switch has 40 interfaces (interfaces 1/1 - 1/16 for eth0 of each blade, 2/1 - 2/16 for eth1, and then 1/17-1/20 and 2/17-2/20 for uplinks). However, there seems to be a sampling limit on the firmware that only allows 32, so that's why I did 1/g1-1/g16 in the command above instead of 1/g1-1/g20.
Also, if you happen to use rancid to manage your dell configs, you will need to modify it to ignore the timeout, as it descreases all the time, so rancid sees it as a config change and will alert every time.
that change goes into "drancid" file in the WriteTerm sub, like so
# ignore sflow countdown timer
if (/(sflow\s+\d+.*timeout)/) {
ProcessHistory("","","","$1 4294967295\n"); next;
}
One other last caveat - there seems to be a bug with the switch not persisting polling and sampling configs after a reboot (I have an open case with Dell on it), so be aware that you may need to re-apply the polling/sampling lines above to get it to work again. It might be specific to stacked switches.
Hope this helps!
also the mandatory timeout is a bitch anyhow, next thing they will include a counter that will reset to factory default or something ;-)
ReplyDeleteMy RANCID was also spammed with changes until I noticed
The above issues were resolved in latest 4.2.1.3 build.
ReplyDelete(i) No Timeout option is now available
(ii) The user was entering the wrong syntax. Ranges are specified in the following manner:
List of ports: sflow 2 polling te1/0/1,te1/0/3
Range of ports: sflow 3 polling te1/0/1-3
Hi all,
ReplyDelete"One other last caveat - there seems to be a bug with the switch not persisting polling and sampling configs after a reboot (I have an open case with Dell on it), so be aware that you may need to re-apply the polling/sampling lines above to get it to work again"
I have the same issue.
It is happening on my PowerConnect 7048 (stacked), 7048R-RA single unit.
The switches were factory shipped with firmware version 4.2.2.3.
I upgraded to firmware version 5.0.1.3 and then configured the switches for management, to discover this problem.
I have reverted to version 4.2.2.3 to realize the problem persists.
Currently have a support case with DELL
Update on my problem:
ReplyDeleteDELL Technical support identifies that there is a bug in their firmware (PCT6248, 7048, 8132) that would\should be fixed in the next release.
creating receiver, samplers and poller via CLI is the workaround as these settings are saving
Hi,
ReplyDeleteI am looking for a workaround to enable sFlow on Dell 5548 and 6240 switches
Unfortunately, I don't have either of those switches to try. Based on previous comments, it sounds like the CLI syntax has changed. Did you look through the updated examples in the comments? Please post the correct configuration settings when you find them.
Delete