Eugene's questions

I am deploying Yealink IP phones (T40G, T23G) on following newtork equipment:
CCR1009-7G-1C-1S+ as a router
CRS328-24P-4S+ as switches

Config of the switch in trouble is here.

I set up DHCP option 132 to configure VLAN for phones.

Before deploying such config in production, it was proved working on hAP AC^2. However while deploying it on site I encountered strange bug, which looks like this.

A phone successfully gets IP from untagged VLAN with option 132. Then it releases this IP and requests a new one from voice VLAN. The DHCP server assigns this new IP and sends "ACK" message, which is never received by the phone.

The problem seems to be switch-related. Here (careful, it's in Russian!) an admin managed to overcome this problem by disabling VLAN MAC learning. However this option is valid for SwitchOS, while my switches run RouterOS.

Also, this reddit thread has a brief summary of what needs to be configured, but it is quite outdated (5 yrs today).

This is what sniffing at phone port shows.
Notice reply packet at 49.348 from 192.168.10.40 and its absence at 50.0 subnet:

Switches had 6.48.1 stable ROS. I tried long-term 6.47.9, it didn't help. Switch in question has its own VLAN 50 IP address, which can be pinged from the router. Firewall is disabled. Phone is updated with latest firmware.

What should I do to find the cause and resolve the problem?

I am deploying a solution using IKEv2+ipsec with certificates to connect roadwarriors to corporate network. Mikrotik CHR is used as entry point.

All was swift until I started deploying the solution on Dell notebooks. Once connection to the router is established, laptop doesn't get split includes, and only VPN subnet is available. On the contrary, my admin PC, which is stationary workstation, has no such problems.

Windows 10 receiving split includes using DHCP. After some research I found out that for some reason, Dell-provided Windows 10 Pro 1909 fails to send DHCP request to the router. Laptops get their address, DNS, only split include routes are lost. Also, DHCP works well on Wi-Fi adapter.

What was done:

• Logs at router were examined for both laptop and admin PC. No DHCP requests was found when laptop connects.
• Traffic was sniffed at Microtik CHR: DHCP request, which comes from admin machine, doesn't come from notebook.
• Traffic was sniffed at notebook, and no DHCP requests were detected.

Rebooting, resetting ip and winsock using netsh, reverting to older wi-fi driver, deleting and re-creating WAN Miniports, enforcing DHCP for a connection, dancing around a laptop - all that didn't help.

Currently the only solution that works is a clean MSDN version of Windows 10 1909 installation. With this one, laptops get their split includes well. However, it doesn't seem a sound solution to me.

My questions are:

• What is possible cause of the problem?
• What can be done to fix it?

I am deploying some 50 of SIP-T46S phones, and got stuck with really weird behavior.

According to Yealink manuals, "new" provisioning process is getting .boot file and applying configs in order specified in there.

I decided to go this way: y000000000000000.boot -> common.cfg -> mac-specific.cfg

y000000000000000.boot contains this:

[T46S]include:config <T46S/$MAC.cfg>
include:config <common.cfg>

overwrite_mode = 1
specific_model.excluded_mode=0

Now I can read from TFTP logs that phones successfully download both .cfgs, but for some reason doesn't apply settings from common.cfg. Settings from $MAC.cfg are applied successfully.

Switching lines doesn't help. When I move lines from common.cfg to $MAC.cfg, they are applied.

Firmware Version 66.84.0.15
Hardware Version 66.0.0.128.0.0.0

The question is: why non mac-specific config is dropped?

Please help to understand and solve the problem.
Many thanks!

There is some unexpected behaviour of Windows SMB shares.

If a share is named all-uppercase, then it changes case depending on how you enter server name.

Hence if there is an example share named "SHARE", then entering \\SERVER in Windows Explorer results in share name appearing as "share", and entering \\server or \\Server results in "SHARE".

At the same time, shares with both upper- and lowercase characters in name are presented as they are named, so "Share" will remain "Share" in both \\SERVER and \\server (\\Server) cases.

It looks like the problem is in SMB protocol, as all windows versions seem to have this problem. In my case there is Windows Server 2016, and shares get listed from Windows Server 2016 and Windows 7 computers.

What could be the solution for this problem? (i.g., how to make SMB display names properly regardless of server name?)

I work for a small company with ~180 users. We are planning a complete overhaul of our infrastrucure.

My question is whether Ubiquity equipment is suitable for our task (which is explained below), or we should look for other vendors.

The basic idea is building an hyperconverged WS 2016 DC cluster of 3 servers and connecting them via 10GbE links for SAN and LAN. This would require a couple of good 10GbE switches (mostly for low latency).

Also, we plan rebuilding LAN basing on a star design. I would build this all on two 10GbE switches, serving as an aggregation level switches for LAN, and as an access level switches for SAN, backing each other up. LAN and SAN will be segregated by VLANs.

Thus, these switches should also be able to route VLANs, so probably L3 is required (AFAIK L2+ cannot route VLANs).

My current choice:

• US-16XG / ES-16XG for aggregation level
• US-48-750W for access level (+IP phones)

Please advise, should I use them or decline this idea. Thank you!

I'm setting a test lab to evaluate best solution for future production use. The production farm intended for an SMB, so budget is present, but it is also limited.

Goal for production: 3 hyperconverged servers with Windows Server 2012 R2 failover virtualization cluster and Software Defined Storage solution as shared storage. In short term the cluster will be expanded to total of 5 servers. SAN network is dedicated.

Goal for test lab: find an SDS solution, that meets following criteria:
1. Provides shared storage for Hyper-V cluster.
2. Scale-Out: can add-remove disks (and preferably - nodes as well) without shutting down the cluster.
3. Fault-tolerant. Available after loss of 1 node (if recoverable after losing 2 nodes - it is great!).
4. Low network overhead/latency (will be used for SQL Server as well).
5. Have reasonable pricing (Storage Spaces Direct is not acceptable for this reason).

After reading and looking at some products, my short-list is EMC ScaleIO and Starwind Virtual SAN.

I tried them both, and found that Starwind VSAN provides quite limited HA: as I understood it, this solution only mirrors virtual disk (namely, a file) across nodes, and only allows expanding capacity within limits of hosting disk. ScaleIO, on the contrary, spreads data across nodes, and allows adding new storage and rebalancing volumes.

So, my questions are:

• Are my assumptions correct, or Starwind VSAN allows creating an HA volume across several disks in every node and add disks later?
• What solution is better for my application in your opinion (please explain)?
• What are drawbacks of advised solution?

Thank you in advance!