v1.4.0rc1-16 Bug reports and comments

[IntL-Daniel]

No ladybugs there nor other bugs :-)

[designweb]

going back to my switch that had the fan problem where you guys were discussing it not being reported properly, I took it down to 1.3.9 and now up to 1.4.rc12 and it is still reporting fan missing but at least now two fans are shown running on the status page and the speeds are varying properly with temperature

Jan 1 00:00:43 system: starting ntpclient
Apr 16 03:54:31 dropbear[1416]: Running in background
Apr 16 03:54:40 switch[1491]: Detected warm boot
Apr 16 03:57:00 switch[1434]: Warning: expected 2 fans but found 1
Apr 16 03:57:00 sSMTP[1753]: RCPT TO: (550 Sender verify failed)

[sirhc]

going back to my switch that had the fan problem where you guys were discussing it not being reported properly, I took it down to 1.3.9 and now up to 1.4.rc12 and it is still reporting fan missing but at least now two fans are shown running on the status page and the speeds are varying properly with temperature

Jan 1 00:00:43 system: starting ntpclient
Apr 16 03:54:31 dropbear[1416]: Running in background
Apr 16 03:54:40 switch[1491]: Detected warm boot
Apr 16 03:57:00 switch[1434]: Warning: expected 2 fans but found 1
Apr 16 03:57:00 sSMTP[1753]: RCPT TO: (550 Sender verify failed)

If you reboot it does it report this every time?

If so then maybe you should RMA it?

[IntL-Daniel]

For me...no more false voltage warnings on RC12 (almost 48h uptime) :). I would only recommend:

* fix Web UI for mobile - imposible to select options per port (bounce, reset counters, port detrails..) - tested on Firefox for Android
* change "root" sender of email notification by sender's email
* change copyright at botom to 2016 maybe :)

[wisp-wand]

Hi Guys, RC12 looked good on the bench, went and installed it back onsite today, but alas same old issue (see below log)

Can I ask for one feature that will make this switch usable (for us) ?
The ability to stop the software reacting to thermal/voltage out of specification

Its clear the i2c bus is getting corrupt info for whatever reason, but if the switch simply could be set to ignore the info, then we could use the other 95% of the features this switch has and turn off the 5% that cause the problem.
Be interested in your comments.

Apr 23 14:39:33 switch[805]: Disabling PoE power on port 2 due to Power Supply Thermal out of specification (controller is 36 C)
Apr 23 14:39:34 kernel: link state changed to 'down' on port 2
Apr 23 14:39:38 switch[805]: Enabling PoE power (24V) on port 2 due to Power Supply Thermal within specification
Apr 23 14:39:41 kernel: link state changed to 'up' on port 2
Apr 23 14:39:45 kernel: link state changed to 'down' on port 2
Apr 23 14:39:47 kernel: link state changed to 'up' on port 2
Apr 23 14:40:04 kernel: link state changed to 'down' on port 2
Apr 23 14:40:06 kernel: link state changed to 'up' on port 2
Apr 23 14:40:09 kernel: link state changed to 'down' on port 2
Apr 23 14:40:11 kernel: link state changed to 'up' on port 2

[sirhc]

We are working on this issue.
It does not affect everyone, just a select few and we are not sure why, guess your just lucky.
I am sending you a PM to a link to rc15 if you want to try it.
If you try it please report back here.

You do realize the DC switches do NOT work at or near FM broadcast sites?
Discussed in this thread: http://forum.netonix.com/viewtopic.php?f=17&t=1342

However we have been trying to make the newer firmware better with FM tower locations or near one anyway but we will never be able to fully resolve the issue in firmware as it required hardware changes.

We "may" consider a firmware feature like you mention that disables the SMART capabilities of the WS-12-250-DC and WS-8-250-DC which would allow the older hardware to function at these locations but in a dumb mode.

There would be no monitoring of the DC input voltages, temperature of power supply, and such. WOuld also disable a lot of functions on the Power Tab.

[wisp-wand]

I got your PM and will update the switch. Unfortunately it failed completely again early hours of this morning so will need another site visit to recover it, before I can load RC15 on.

I don't know much about your software, but there would be no reason to disable all i2c - just put a trap in so that as the switch reacts to a value, check to see if the 'react to i2c value' = yes. Doesn't even need to be in the GUI.

I still think this issue would be better solved by averaging the i2c values rather than taking one reading and reacting to it.

That might take more code though, and I guess if you have such a minority of users affected it might not work out commercially.

I think it might be better to clarify that this issue is not just at FM sites. This switch is not located at an FM site. It has no transmitters nearby, it has a substation about 500 metres away and a train track (electric) about 300 metres away the issue is most likely related to EMC/EMF but not always at FM sites.

Wayne

[sirhc]

We do average the values and we are working on the code as best we can within the hardware limitations of the current design to deal with sites that have some sort of interference messing with the I2C.

As I have said we envisioned this switch being used at off grid locations so we did not design in any filtering to deal with interference with the I2C communications. Since this problem has been identified by a "small" number of users using them near FM tower sites or some other type of interference we have gone back and spent a considerable amount of time and money to make design changes that should eliminate this. The new designs are in production now and should hit the market by end of May.

Obviously this location must have some sort of interference that is affecting the I2C communications between the switch and the internal power supply from the sub station or the train tracks. It vould also be getting noise from a bad charger you may have?

We are trying to do as much as possible in software to help but the hardware solutions is what is really needed for sites like this.

Can you use an AC powered switch at this location for now and use that switch at another location away from possible interference?

Turning off the I2C communications may not be a good idea but we are investigating offering that option.

Another possible problem can be bad or incompatible SFP modules as they share the same I2C bus.

[sporkman]

I just noticed a switch running at 100% cpu, (ws-6-mini) and it was on rc10, so I upgraded to rc12, but same result.

See below from top:

Code: Select all

 
top - 22:06:06 up 40 min, 0 users, load average: 1.39, 1.27, 1.09
Tasks: 64 total, 3 running, 61 sleeping, 0 stopped, 0 zombie
Cpu(s): 93.9%us, 5.4%sy, 0.0%ni, 0.0%id, 0.0%wa, 0.0%hi, 0.6%si, 0.0%st
Mem: 126348k total, 50388k used, 75960k free, 4468k buffers
Swap: 0k total, 0k used, 0k free, 16372k cached
 PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
1639 bwayadmi 20 0 4888 2964 552 R 87.3 2.3 34:22.87 discovery probe
1631 bwayadmi 20 0 4888 2964 552 S 3.6 2.3 1:19.02 switch
1641 bwayadmi RT 0 4888 2964 552 S 2.0 2.3 0:42.88 i2c thread
2036 bwayadmi 20 0 2116 632 452 R 1.6 0.5 0:00.19 top
1336 bwayadmi 20 0 6944 1940 1052 S 1.6 1.5 0:02.03 lighttpd

 

In the last revision, the "switch" process was pegging the cpu, now it's the "discovery probe" process.

[sirhc]

FIXED
- Fixed SNMP reported uptime - rc2
- Fixed rare problem with mac table not loading - rc3
- Fixed small UI issues with Firefox and IE - rc3
- Fixed SNMP CPU counters - rc4
- Fixed SNMP firmware version string - rc4
- Fixed so log messages now used VLAN ID instead of index - rc4
- Fixed RADIUS to work with SSH- rc9
- Fixed port not being in VLAN Bug- rc9
- Fixed pps being included in totals when port was excluded- rc9
- Fixed false fan failure detection *we hope*- rc9
- Fixed cleared log to indicate who cleared it- rc9
- Fixed so no duplicate VLAN ID allowed- rc9
- Fixed SSH session ending messages so not reported by dropbear- rc9
- Fixed Firefox autocomplete (workaround) - rc10
- Fixed language encoding - rc10
- Fixed console in UI - rc10
- Fixed graphs on Status Tab - rc11
- Fixed some incorrect voltage readings/alerts - rc11
- Fixed some more I2C communications problems mostly for SMART DC models. - rc12
- Fixed RSTP path cost now able to set to 1 - rc12
- Fixed Memory leak - rc16
- Fixed False fan alerts - rc16
- Fixed more I2C communications problems all models - rc16

ENHANCEMENTS
- Increased MAX allowed DC input voltage for WS-8-250-DC and WS-12-250-DC to 61.5V for off grid solar sites - rc2
- IMPROVED SMART DC switches I2C communication- rc9
- Added HTTP Port to be changed- rc9
- Added telnet for CLI access- rc9
- Added Discovery to Tools menu- rc9
- Added multiple languages support- rc9
- Added VLAN information to SNMP- rc9
- Added Revert Timer Help Balloon- rc9
- Added total power consumption to SNMP- rc9
- Added DC input voltage graph on the Device/Status Tab- rc9
- Added tooltips to buttons- rc9
- Added increased SMTP Username length- rc9
- Added Power tab on all models for time conditions- rc9
- Change ifDescr and ifAlias- rc9
- Time Condition graphic on Christmas tree rc10
- Added To line to email header rc10
- Added more graphs for SMART DC models under Device/Status Tab - rc11
- Added an Enable/Disable check box for TFTP backup - rc11
- Reduced CPU usage - rc12

KNOWN ISSUES
We are currently investigating a problem reported by a few customers as discussed in this thread: viewtopic.php?f=17&t=1654

At this time we have not identified the cause of this issue nor are we certain if this is our problem or not and is not fixed in rc16.

We feel it may have something to do with Flow Control and possibly involves RSTP?

If you poll the switch too fast with SNMP you can cause excessive CPU strain and possibly cause the switch to lock up.

rc16 Released 5/3/2016 - If your running v1.4.0rcX or have a SMART DC Switch I recommend upgrading
rc12 Released 4/14/2016 - If your running v1.4.0rcX or have a SMART DC Switch I recommend upgrading
rc11 Released 4/12/2016
rc10 Released 4/08/2016
rc9 Released 4/05/2016
rc4 Released 2/29/2016
rc3 Released 2/11/2016
rc2 Released 2/06/2016