v1.5.25 Bug Reports and Comments

[mayheart]

Is anyone else noticing an issue where applying a new configuration causes a brief forwarding hit? In my case I was enabling ports and changing descriptions, and 2 of my IGP links (OSPF /w BFD - 5x200ms) that pass through completely different ports (not changed) reset.

Couldn't say whether its a new problem with this new software, but its the first time I am noticing it.

Netonix switches have always done this. It gets worse the more vlans you have configured.

That being said, 1.5.25 improves this for us greatly, it's MUCH faster at applying changes. WS3 line up is even faster for this. I don't see BFD drop at all.

[RebusCom]

We again were hit with another mass cold reboot thought to be associated with RSTP. Five WS switches comprising a common RSTP network at different tower sites underwent a spontaneous simultaneous cold reboot, taking down all the POE radios in the process. An interesting twist this time though is a sixth switch co-located at one of the sites also participated in the cold reboot, though it's not part of the actual RSTP chain (RSTP disabled) and none are on an external watchdog. It does have loop protection enabled, as do all the others. Remaining switches not co-located or part of RSTP did not reboot and there are no other co-located switches. Non-POE equipment, e.g., routers, UPS, environmental monitors, at the sites were unaffected.

Following this event, once the network came back up the switches all failed to get NTP. For the next 24 hours there were multiple instances of
admin: stopped ntp daemon
admin: started ntp daemon

admin: sync time via ntp

Exactly 24 hours plus two minutes later the sync time via ntp succeeds on all the switches that rebooted.

All switches are running v1.5.25, which has not fixed the problems with random mass reboots or entirely fixed the NTP issue from previous versions going back to at least v1.5.8 when we experienced our first occurrence. I've reported previous events. Fortunately the events are infrequent but it's highly disruptive (and nerve wracking) when it occurs.

In addition, following this event, the Netonix Manager v1.0.24 shows incorrect run times for the affected switches, displaying run times as though the event never occurred though the run time is correct on each switch GUI. A workaround for this is to wait the 24 hours, then remove the switch from monitoring and add it back in. Then the correct run time is displayed. It appears the manager does not get it's runtime from the switches except at initialization of monitoring, but then relies on it's own clock from the device it's running on.

[sirhc]

If using snmp how often being read.

Only 1 snmp server?

Are you using more then one monitors?

Try disabling all non critical.services one at a time like discovery protocols, snmp, email.....

Are you using access control list?

Do you disable RSTP on ports that is impossible for a loop. For instance i disable ports with APs and my towers have an in and out backhaul link creating OSPF ring so i disable RSTP on those ports as it will cause delay in convergences.

Are you using a power management strip that allows you to reboot equipment?

Not saying you dont have an issue but I never see this at my WISP and its not a wide spread issue so something is cauing it. Maybe our issue maybe something your using or in a way your using it?

If you could do a screen shot of ALL tabs with changes from default, upload images to post using attach file link below blue submit post button. Do not link images to third party sites.

Only redact what is needed like VALID IP ADDRESSES, EMAIL INFO, SNMP INFO.... Never understand why people redact things like MAC or non routable IPs like 172.17.X.X, or 192.168.X.X, or 10.X.X.X?

Is this a flat network or routed? If a flat network and you have a bad piece of equipment that spews garbage on network can cause shit to go bad.

For instance my office building would go dark then come back. Got worse and worse. Tracked it down EVENTUALLY to a bad 6 port dumb switch that sat up behind cash register so we could plug in things like printers and such where there was one ethernet wall jack. Replaced switch problem gonr. It was an OLD SMC dump switch.

Had similar issue many years ago before netonix turned out to be a bad media converter. Eventually got so bad net stayed down until removed. If we plugged it in on any network it dropped it. Devices was spewing 1g line speed of malformed garbage packets.