1.3.7 - Odd Packet Loss Issue

[mhoppes]

I have a network which was running (I believe 1.3.3?) and was working fine. Now, after loading 1.3.7 on all of the switches I am experiencing around 1% packet loss, even going from an EdgeRouter to the Netonix!!! I'm not sure what to make of this?

I'm running a further battery of tests but curious if anyone else has seen this.

[sirhc]

Nothing changed between those versions to the switch core configuration. In fact nothing has changed in the switch core configuration since v1.1.8

Maybe try rebooting the radios, check all interface for errors.

Work your way out from the front to the back.

[mhoppes]

That's what I'm doing now. The odd thing is I'm seeing the packet loss on the interface directly between the Netonix and the EdgeRouter.

[mhoppes]

After playing around with this for about an hour this evening, I was confident it was an issue between the EdgeRouter and our switch (I didn't even have to go out into the network to see the packet loss).

So, I rolled the switch connected to the router back to 1.3.2 and everything seems fine. Correction on the version. I was upgrading to 1.3.8. Rolling back to 1.3.2 seems to have corrected the packet loss issues.

[mhoppes]

Had a second network segment where I was getting massive packet loss (sometimes as high as 14%). Had to roll all three Netonix switches in the chain back to 1.3.2 in order to get the packet loss to stop. I've been watching the one switch and packet path now for a good 25 - 30 minutes and the packet loss is completely gone since downgrading.

[sirhc]

Why would you be upgrading to v1.3.7 instead of v1.3.8 and to be honest I would use v1.3.9rc13

Now I am not sure if your traffic is utilizing any SFP ports but if so you may need to try both the 1G or AUTO to determine which mode works best with your specific SFP module under the ports tab for Speed/Duplex for that port which is why we added that ability lately for SFP ports. Now you also have to try each mode with each mode on the device the SFP port connects to and I would be a little afraid to try this remotely as 1 mode may not work with your SFP module properly or in that mode may not talk to the other end properly.

Other than that I can say that I have used all versions and no change between v1.1.8 and v1.1.3.9rc13 would cause packet loss except maybe SFP module. You need to look at little deeper into this Matt

Another "possible" thing was recently a fix was put in place where the switch did not deal with domain names for some configurations in the Device/Configuration Tab such as syslog, TFTP, SNMP servers which would cause certain daemons to eat up CPU time which "could" possibly cause an issue.

But like I said v1.3.9rc13 is what I would use as these were fixed but "definitely" I would not use v1.3.7

Now all these above issues were discussed in the release notes and or the release comments thread.

[mhoppes]

Title is wrong. Upgraded to 1.3.8 not .7.

No SFPs being used. But defy she packet loss on .8 that is not there on .2.

I can try the release candidate and let you know.

[mhoppes]

Didn't try release candidate, but upgraded back to 1.3.8 and there is no issue. The issue was the result of a RADWin radio in the mix that no one was aware of (ugh!)

[sirhc]

Didn't try release candidate, but upgraded back to 1.3.8 and there is no issue. The issue was the result of a RADWin radio in the mix that no one was aware of (ugh!)

Thanks for causing a worldwide panic, and then thanks for quelling that panic!

[mhoppes]

It was very coincidental.... but had nothing to do with the firmware upgrades.... That's why I updated the thread! Didn't want to keep a panic going!