Am I using the right switch (SG350X-48-K9) ?

[FHausH]

I am representing a school with 4 computer rooms each with 20 students + 1 teacher + 1 printer (Windows 7 and Win-Domain).

So 88 devices are connected to 2 stacked Cisco SG350X-48-K9 (48x RJ-45, 2x RJ-45/SFP+, 2x SFP+) + Windows-Server + Internet.

We have a software for the teacher that can monitor each student, distribute files etc.

Each computer is  connected via a CAT7 cable.

Our problem:

When all 4 rooms are in use after a while the switch ist not accepting input any more, server connections break down and we need to restart EVERYTHING - server + switches + student computers.

This may work the rest of the day, but we may have another crash the same day.

Big question:

Did we purchase the wrong switch?

The 2 switches are connected via 2 CAT7 cables and are on the latest firmware release.

and

What can we do to find out what exacly goes wrong.

Any help highly welcome!

[Seb Rupik]

Hi there,

Can you share the running config of your switch stack?

Are you sending the switch logs to a syslog server, can you share them? If not have you had a device attached to console port capturing the console output during one of your outages? What did it show?

cheers,

Seb.

[FHausH]

Many thanks for the quick response.

I am attaching the Tech Support file.

No currently no syslog server.

And no device for the console.

If required, I will work on that.

By the way:

I did not find instructions on the proper cabling of the stacking cables.

Currently they are XG -> XG2 on both switches

[Seb Rupik]

Your stack configuration is correct. The 350X can only use the TE ports, and upon detecting a compliant switch will bundle the links into a LAG.

You config also looks fine to me. The only adjustment I would make is to move your user VLAN off of VLAN1 to some other number, as VLAN1 is used by control traffic. Although unlikely to be the cause of your issue in this instance it is best practice and should be followed if you expand your network.

I think the next step is to capture the logs during a failure. If the failure is fairly easy to anticipate I suggest attaching the console cable, increasing the buffer size on the terminal and just capturing the messages. This should give us the clue we need. The switch is already logging at level-7 (info) so there is nothing to adjust.

cheers,

Seb.

[FHausH]

Hi Seb.!
Many thanks for your support.
I have now used a qnap nas to act as a syslog server.
I am not sure what you mean by: "The only adjustment I would make is to move your user VLAN off of VLAN1 to some other number". No ports are assigned to any VLAN . I have created a VLAN 2 but then I did not know, what to do next, any action seemed to dangerous to me.
I will post again, when it happens next and send the logfile.

[FHausH]

Today we had no problem, but this is also because my colleagues reboot the switch every morning preventivly, because the computer lessons are very much disturbed, if the switch goes down.

I have attached a typica log. It shows the permanent + random up and down of the connections to the computers.

In 3 minutes over 130 log entries.

This is typical and starts, when the computers are swiched on and stops, when the last computer is switched off.

I have attached a video too.

The video shows the LEDs with NO COMPUTER switched on.

Other switches show just a steady light but here they are flickering like a christmas tree.

Any ideas welcome. Thanks

[mipopov]

Hello,

Why you set on the second unit all the gigabit ports in speed 100?

And is it also set manually to 100Mbits on the end devices connected to these ports?

[FHausH]

Thanks for the information.

A colleague tried to eliminate the up/down messages by setting the connection speed manually.

Resetting was done correctly only on Unit 1 and not on unit 2.

I have corrected this and attached the latest diagnostics file.

We cannot give any more feedback, because there is school holidays till Jan 8.

After this I will report.

[viningele]

Did you select both ports used in the stack set up page for both switches?

If you’re using two CAT-7s between two switches shouldn’t that be a ring topology not a chain?  If wired for ringg and the switches think its a chain could it be creating broadcast storms if STP sn’t config’d. 

https://youtu.be/VIvHgH7eNgc

[mipopov]

2 switches in stack cannot create Ring topology - it is only Chain

[FHausH]

Thanks for the contributions to my issue.

So a question:

Should I remove one of the cables?

If yes: Is there a preference in what stacking cable goes where?

[viningele]

FHausH

, watch the video in my earlier post and in stack management just make sure you designate which ever port you use as a stack port.  If you haven't already do both port as currently wired and see if that resolves your problem.  If they are just remove one cable and undesignate which ever port you choose on both switches and see if that solves your problem.  FYI, I'm no expert either.  Supposedly from the video you can designate 4 ports I think so I think so maybe they form a LAG?  All I know is a redundant loop from one switch to another that's not configure properly can cause a broadcast storm that will easily crash a network, sometimes immediately sometimes only when a certain chatty client comes online.

[viningele]

From the manual: “To stack two or more devices, reconfigure the desired network ports as stack ports in the devices and connect the devices with the resulting stack ports in a ring or chain topology.”

based on the above  I would think two switches could be stacked in a ring topology.

Manual, stacking starts at page 116.

.https://www.cisco.com/c/dam/en/us/td/docs/switches/lan/csbms/350xg/admin_guide/AG_Tesla_350_550.pdf

Page 126 Stack Ports:

“The following ports can be stack ports:
• XG Devices—All ports can be stack ports
• X Devices—The four XG uplink ports can be stack ports”

Since you have an X model not XG only four ports are stackable and I would think only they could be assigned in the stack management web gui page.

[viningele]

Looking at the differences between the X and XG models, ports on the X are 1 gig whereas port on the XG are 10gig which is why they are all stackable on the XG models.  Looks like you need to stack on the SFP+ combo or SFP+ ports.  

That would be my guess anyway.  So if you have 2 regular ports connectected and not set up for LAG you’ve created a network loop so for now disco one.  Then get the modules needed for the 10g stack, read the manual and watch the video I posted earlier.  

[FHausH]

Many thanks for the advice.
I have considered this, and the cables are in the right place.
I have removed one cable so there is only one stacking cable left.
I need to wait until the students are back next week. Then we will see.