This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.

You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

NPM 10 - Procurve Transmit Discards

LukeC64 over 13 years ago

Hi,

I am Luke Cameron, I am a Senior Network Engineer for my company. We are trying NPM on our network with our Procurve and Cisco switches.

We have noticed NPM is reporting loads of Transmit Discards on the Procurve Switches the links are not being used that heavy some only 10% also have the latest firmware on them.

I have looked at the switch Stat's and its not reporting any issues.

Is this a bug with NPM.

Switch is a 2610/12PWR running Firmware R.11.54 and before that R.11.30 and it was doing the same thing with the older firmware on there.

Top Replies

FormerMember over 13 years ago in reply to FormerMember +1

Bogus transmit discards counters have been a common problem on the ProCurve 2600 and 2800 family of switches. I've had switches that weren't connected to anything but a single idle PC and the error counts…

0 byrona over 13 years ago

Luke, welcome to the Thwack forums!
We use Procurve switches here as well and I don't see this problem (I just went and checked to make sure). Orion is just telling you the information that it is getting from the switches.
I am not the strongest networking guy in the world but you might check to make sure there aren't any duplex mis-matches.
Cancel
Vote Up 0 Vote Down

Cancel
0 FormerMember over 13 years ago

I don't have the problem either.
Cancel
Vote Up 0 Vote Down

Cancel
0 FormerMember over 13 years ago in reply to FormerMember

Bogus transmit discards counters have been a common problem on the ProCurve 2600 and 2800 family of switches. I've had switches that weren't connected to anything but a single idle PC and the error counts were continually incrementing, even when there was no data. The problem would disappear with a new release of code and then return with a future release.
I used to test new releases and if the counter problem was there, I would stay on the older release. In one case, I was running a version that was a couple years old to avoid the bad error count problems.
A good way to test to see if it is the counter bug and not a real problem is to setup only two devices and ftp (or another TCP based file copy) through the switch and between devices. Watch the conversation with an analyzer. If the error counters are incrementing on the switch, but there are no packet drops (retransmisisons) seen on the analyzer, then the counters are in error.
Now, if you see the drop counter increment on the switch and retransmissions on the analyzer, then you have real problem.
If you have a duplexing problem, the switch should show that in the event log and you'll also see the collision counter increment on that port.
Rick
Cancel
Vote Up +1 Vote Down

Cancel
0 bshopp over 13 years ago in reply to FormerMember

Here are the OID's we use to gather this
Interface Errors and Discards:
OID = ifInDiscards
1.3.6.1.2.1.2.2.1.13.
OID = ifInErrors
1.3.6.1.2.1.2.2.1.14.
OID = ifOutDiscards
1.3.6.1.2.1.2.2.1.19.
OID = ifOutErrors
1.3.6.1.2.1.2.2.1.20.
Cancel
Vote Up 0 Vote Down

Cancel

OID = ifInDiscards	1.3.6.1.2.1.2.2.1.13.
OID = ifInErrors	1.3.6.1.2.1.2.2.1.14.
OID = ifOutDiscards	1.3.6.1.2.1.2.2.1.19.
OID = ifOutErrors	1.3.6.1.2.1.2.2.1.20.