v1.5.21 Bug Reports and Comments

[mayheart]

I ran into an issue with upgrading from 1.5.14 to 1.5.21, username/password did not get reset for the webui. I was unable to login using the old login/password or admin/admin. SSH was fine though.

After using credentials commands in SSH it fixed its self. I did make a backup of the config before and after the password change.

As for PPPoE, it seems to be working fine for me but I have my ports set to 9000 MTU. I use mostly FS.COM Cisco branded SFPs.

[sirhc]

I ran into an issue with upgrading from 1.5.14 to 1.5.21, username/password did not get reset for the webui.

I was unable to login using the old login/password or admin/admin. SSH was fine though.

After using credentials commands in SSH it fixed its self. I did make a backup of the config before and after the password change.

The user name does not get reset, only the password gets defaulted to "admin" do to upgrading security hash, the password stays the same.

Weird?

Either way glad you got in via SSH and recovered.

As for PPPoE, it seems to be working fine for me but I have my ports set to 9000 MTU. I use mostly FS.COM Cisco branded SFPs.

GOOD STATEMENT

I am guessing MTU is playing a part in a lot of peoples problems. Most people don't account for the extra bytes needed for things like VLANs, PPPoE and assume you can use 1500 even when just using VLANs which is incorrect. There are extensive conversations delving deep into this on this forum. This is why we default to an MTU 1528 as most people use VLANs. Then that does not necessarily include QinQ or Double Tagged.

You are correct PPPoE adds extra bytes to needed MTU.

This is a GOOD article to read. https://www.networkworld.com/article/74 ... ssues.html

Remember if using VLANs, Multicast, PPPoE, among other encapsulating protocols make sure you take this into account then make sure when you get to your EDGE and enter the WWW your back down to 1500 MTU as the WWW standard MTU is 1500.

[Flo]

I ran into an issue with upgrading from 1.5.14 to 1.5.21, username/password did not get reset for the webui. I was unable to login using the old login/password or admin/admin. SSH was fine though.

After using credentials commands in SSH it fixed its self. I did make a backup of the config before and after the password change.

As for PPPoE, it seems to be working fine for me but I have my ports set to 9000 MTU. I use mostly FS.COM Cisco branded SFPs.

That's interesting. We also use FS.com SFPs but it does not work after upgrading to v1.5.21. PPPoE servers are MikroTik CCRs 2xxx.
Tried both: Standard MTU 1528 and increasing the MTU to 1556. My guess is that the MTU setting only gets pushed to the SFP module after a hard reset (unplug / re-insert).

I have ordered a few other SFP modules from the compatible list and will give it a retry the next days.

[tcknudson]

I just upgraded a WS-12-250AC switch to 1.5.21, went right from 1.5.16 to avoid bricking it.

Bricked it anyway. POE lights come on as programmed but will not pass any communication.
This is getting problematic as I do not have any more replacements?

I reset the switch to factory specs, and it allowed to program from scratch, so it did allow the upgrade,

[sirhc]

I just upgraded a WS-12-250AC switch to 1.5.21, went right from 1.5.16 to avoid bricking it.

Bricked it anyway. POE lights come on as programmed but will not pass any communication.
This is getting problematic as I do not have any more replacements?

Sorry for your issue.
If the unit had been up for a LONG time as there was a memory leak a reboot first might be good idea but normally not needed.

Hook up console see what you see.

A power on factory default might help.
Or possibly a console recovery (see post on instructions)

I did have some issues with previous versions from v1.5.14 or v1.5.16 to v1.5.17 and v1.5.19 but was able to recover them.
One was stuck in reboot loop but stopped after about 4 then loaded, another one required power cycle.
We suspected bad SFPs in the one unit so replaced them upgraded again no issue.

I upgraded 50+ switches to v1.5.21 and had no real issues.

There are literally hundreds of thousands of units out there around the world with all differnt services and configs, I am sure sadly some will run into issues but once up to v1.5.21 these issues should be FAR less.

[mayheart]

I ran into an issue with upgrading from 1.5.14 to 1.5.21, username/password did not get reset for the webui. I was unable to login using the old login/password or admin/admin. SSH was fine though.

After using credentials commands in SSH it fixed its self. I did make a backup of the config before and after the password change.

As for PPPoE, it seems to be working fine for me but I have my ports set to 9000 MTU. I use mostly FS.COM Cisco branded SFPs.

That's interesting. We also use FS.com SFPs but it does not work after upgrading to v1.5.21. PPPoE servers are MikroTik CCRs 2xxx.
Tried both: Standard MTU 1528 and increasing the MTU to 1556. My guess is that the MTU setting only gets pushed to the SFP module after a hard reset (unplug / re-insert).

I have ordered a few other SFP modules from the compatible list and will give it a retry the next days.

I did some bench testing here with a SFP, set the port to 1528. PPPoE worked, was fine after a hard power cycle as well.

The only difference is we're using Cisco ASR9000s configured as a BNG.

[sirhc]

Just after my post on Nov 25 I went a head and upgraded our switches to 1.5.21 (4 total, 1 more had been updated 2 days prior to test but don't have anything to report.). Memory has been rock solid. The trend we were seeing before was the available memory would go down 1-3MB per day. Available Memory has been a flat line so far the last 7 days. UPGRADE!

[RTGLW]

Hi team,

We believe we've found an issue with DHCP on 1.5.21. TLDR: the dhcp service does not appear to run when the switch boots.

Easy fix, well check and put in next release. To be honest that is something we hardly ever use or test so our bad.

Thanks but for now just static assign it I WOULD NOT USE OLDER FIRMWARE

Appreciate the acknowledgment and looking forward to a fix in the next release. Will that come after you guys have addressed the WS3 line or will a "hotfix" be deployed?

We'll continue testing and roll out 1.5.21 to our already statically assigned switches (150+ of these) first. This'll be a good test bed in general for any other hidden issues we may be able to help identify. Unfortunately, we have nearing on 100 DHCP switches and don't have the capacity to make manual changes to these and then subsequently revert those changes when the issue is fixed.

We're beyond thrilled to see these improvements and fixes on the WS platform. Bugs on huge software changes are to be expected, so big kudos to you and Stephen for fixing so many of these quickly as well. We're just looking for it to stabilize more (or be as stable as it can be) to bring all of our switches to parity. We lab test and deploy new FW to our network very carefully, so doing that on an extremely frequent cadence isn't practical for us.

[sirhc]

There will be a v1.5.22 soon.
Stephen is working his ass off but also right now dealing with a sick 16 month old. (Day Care facilities, nothing but a germ fest.)

Fixes planned:
SMTP sender
Graph anomalies
DHCP

{switch-name}# web_conf_check
sh: php-cli: not found

That's interesting. We also use FS.com SFPs but it does not work after upgrading to v1.5.21. PPPoE servers are MikroTik CCRs 2xxx.
Tried both: Standard MTU 1528 and increasing the MTU to 1556. My guess is that the MTU setting only gets pushed to the SFP module after a hard reset (unplug / re-insert).

I have ordered a few other SFP modules from the compatible list and will give it a retry the next days.

To Flo: Interesting, I'll ask Stephen to check that but please report back what you find with other SFP brands. But even if we missed something in Save/Apply config a warm reboot definitly pushes all config to core. Did you try a warm reboot?

[yahel]

Chris - apologies, I've been MIA and just now seeing your reply.
GREAT - I am thrilled that you (think) you know what is the problem with the surges in the graphs.
Just making sure that what you wrote also applies to Voltage graphs (you posted a traffic graph with a surge - yes, we also see these, but don't mind it).
The voltage graphs sometimes show surges of 300-600v, which would be impossible (I'm pretty sure I sent photos).

Anyways - looking forward to 1.5.22 -- can't wait to upgrade and test it.

Hope y'all had a good Thanksgiving and a chance to get some rest.

Appreciating your hard work!

Thanks,

Yahel.