v1.4.0rc1-16 Bug reports and comments

[sirhc]

Is there any point on putting v1.4.0rc on a WS-8-250-DC, it has 1.3.9 and is running fine.

I'm very happy with 1.3.9, this is the first one we have used but have several in the workshop to start rolling these switches out across our network.

Think we have 8 of these switches, some 6 port, several 8 ports and a few 12 ports, but before I send them out I just wanted to see if it is recommended to upgrade to v1.4rc or just leave them with 1.3.9.

I moved your post into the v1.4.0rcX thread as it probably belongs here.

Anyway I would look at the release notes at the top of this thread and decide.

We are coming out with a Management software for those that have a lot of switches to keep track of and upgrade firmware and whereas the software will work with v1.3.9 and below we recommend upgrading to v1.4.0 if using it.

[therealboss]

Just upgraded a WS-8-250-DC to v1.4 but it didn't reboot after the upgrade, switch is still up but have the following in the log.


Jan 1 00:00:05 netonix: 1.4.0rc4 on WS-8-250-DC
Jan 1 00:00:09 kernel: link state changed to 'up' on port 1
Jan 1 00:00:09 kernel: link state changed to 'up' on port 2
Jan 1 00:00:09 kernel: link state changed to 'up' on port 3
Jan 1 00:00:09 kernel: link state changed to 'up' on port 4
Jan 1 00:00:09 kernel: link state changed to 'up' on port 6
Jan 1 00:00:09 system: Setting MAC address from flash configuration: EC:13:B3:51:04:10
Jan 1 00:00:10 kernel: link state changed to 'up' on port 5
Jan 1 00:00:10 kernel: link state changed to 'up' on port 7
Jan 1 00:00:11 system: starting ntpclient
Jan 1 00:00:12 admin: adding lan (eth0) to firewall zone lan
Jan 1 00:00:13 admin: Unable to query power supply
Jan 1 00:00:16 kernel: link state changed to 'down' on port 1
Jan 1 00:00:16 kernel: link state changed to 'down' on port 2
Jan 1 00:00:16 kernel: link state changed to 'down' on port 3
Jan 1 00:00:16 kernel: link state changed to 'down' on port 4
Jan 1 00:00:16 kernel: link state changed to 'down' on port 5
Jan 1 00:00:16 kernel: link state changed to 'down' on port 6
Jan 1 00:00:16 kernel: link state changed to 'down' on port 7
Jan 1 00:00:17 kernel: link state changed to 'up' on port 1
J
an 1 00:00:17 kernel: link state changed to 'up' on port 3
Jan 1 00:00:17 kernel: link state changed to 'up' on port 2
Jan 1 00:00:18 kernel: link state changed to 'up' on port 4
Jan 1 00:00:18 kernel: link state changed to 'up' on port 6
Jan 1 00:00:18 kernel: link state changed to 'up' on port 7
Jan 1 00:00:19 kernel: link state changed to 'up' on port 5
Mar 3 18:55:35 UI: i2c error setting 0x47 12 184
Mar 3 18:55:56 UI: i2c error setting 0x47 14 196
Mar 3 18:55:57 dropbear[899]: Running in background
Mar 3 18:56:13 switch[916]: Error reading temp sensor
Mar 3 18:56:15 switch[971]: Detected warm boot
Mar 3 18:56:16 switch[922]: Unable to read shutdown flag from power supply after 10 tries, still trying...


Maybe I should have tested this on one of the switches in the workshop and not one in production.




Edit
Switch was working 100% on 1.3.9, Uploaded 1.4 and started upgrade. I logged into the AP feeding this tower and noticed the the traffic stop while switch was upgrading. Logged into station at tower that is powered by the WS-8-250-DC and noticed that it had not rebooted but the switch powering it had.

Site is very remote, NO FM towers close by (7 miles away). Site has small wind turbine and 2 solar panels running into 24v battery bank. Voltage on site 24.7v.

Switch has POE 24v switched on portts 1,2,3,4,5 & 6. Port 7 has a Morningstar plugged into it and port 8 is free.

Switch pulling 1.5A (about 38W)

[lligetfa]

I logged into the AP feeding this tower and noticed the the traffic stop while switch was upgrading. Logged into station at tower that is powered by the WS-8-250-DC and noticed that it had not rebooted but the switch powering it had.

That part is expected but the problem reading sensors is not.

A warm boot is expected which disrupts traffic but does not drop the PoE on the ports. Used to be on early firmware builds we would only lose about 3 or 4 pings but the network goes down for longer now.

[therealboss]

Is there a way to get it to do a full reboot and restart or will I have to drive out to the site to restart it?

The reason I thought it had not rebooted is this line in the log.

Mar 3 18:56:16 switch[922]: Unable to read shutdown flag from power supply after 10 tries, still trying...

[sirhc]

You can try and ask it to reboot again but the only type of reboot that can be done is a soft or warm reboot which of course is different than a power cycle or cold boot.

[therealboss]

Rebooted it again but still heve the following issues:-

Board Temp Error
Fan 1 Speed 10000 RPM But its graph is showing 10080
Power Supply Input Current fluctuating between 1.5A & 2.2A (Guess fan is running flat out.



Is it safe to roll back to 1.3.9 or is it best not to roll back?

Eddit
Also notice Throughput graph on switch showing 50Mbps but in Cacti & Rocket both showing 12.5Mbps
Power Supply Heatsink #1 Temp fluctuating between -2 & +3
CPU Temp 34 before upgrade was reading 52

[sirhc]

You need to look at the fan graph again, the left side of graph is 0 the right side is 10000 and in the middle is the actual RPM.

The speed graphs are correct, if you have no port selected it is overall bandwidth but if you select a port it shows bandwidth for that port.

I would like to see you go on site and power off switch for 20 seconds and the power on and see if issue goes away before going back to v1.3.9 and yes you can go back but try a power cycle first just remember to give it a good 20 seconds to completely discharge before power it back up.

And be a good boy scout and be repaired, a console cable, and worst case a spare switch.

[therealboss]

A reboot sorted it out, the GUI was reading fan speed of 10080 but when we got to the site, the fan was off.


If you are working on your own monitoring solution for your switches, we would be willing to test it for you.


Thanks for your help.

[rockhead]

If it is convenient to update your current production one as well i would say go for it for the simple reason that having them all on the same version makes for simpler maintenance going forward.

[wtm]

I am having a weird problem with a DLI POE-24 unit dropping off my network, wondered it maybe it might have been caused by the MAC table bug that you fixed in 1.4.0 RC3 ? Can you explain more of what that bug was ?

My problems is that we are going to a new /23 network, and if the POE-24 is put on the lower /24 portion, and there is anything on the network assigned with the same last 3 digits of the IP address on the upper /24 of that network the POE-24 can not be contacted remotely from outside the network after 30 minutes!
If the there is nothing assigned on the same last 3 digits of the POE-24 on the upper /24 of the network, the POE-24 is fine !

If you are plugged into the Netonix switch locally, there is no problem accessing the POE-24 unit at anytime.