2.9.18 strange bandwidth and VoIP problems

[surfnet]

I just upgraded my network from 2.9.6 to 2.9.18 and some very strange problems have started to occur.

Our network is made up of CM9 5.8 backhaul with Prism2.4 for the clients. No node has more the 50 people.
We use hotspot for authentication and b/w levels. We upload the list of clients to each mikrotik once a day, and use RADIUS as a backup authentication method.

1. We have had a ton of clients call and say they cant get more than 300k download bandwidth. We use simple queues and have 1 meg, 2 meg and 3 meg levels.

when I do the mikrotik bandwidth test, all appears great. I can get 10 megs or more. My pings time are 33ms to the 8th hop out.

2. 1 client uses Lingo VoIP and that day I upgraded his node, the VoIP stopped working completely. That have asked that I check to make sure the ports 5060 and 5065 are not blocked. I have double checked everything and I know I am not blocking those ports. I have not messed with the firewall settings in months.

3. I have a watchdog program that pings each node every minute of every day. Before the upgrade I would have 100% uptime of 95% of the network. Occasionally I would drop a packet here or there. But in gerneral it was perfect. Now I am missing 30 -50 packets each day on all nodes.

any ideas?

[whalen]

@surfnet

I just upgraded my network from 2.9.6 to 2.9.18 and some very strange problems have started to occur.

Our network is made up of CM9 5.8 backhaul with Prism2.4 for the clients. No node has more the 50 people.
We use hotspot for authentication and b/w levels. We upload the list of clients to each mikrotik once a day, and use RADIUS as a backup authentication method.

1. We have had a ton of clients call and say they cant get more than 300k download bandwidth. We use simple queues and have 1 meg, 2 meg and 3 meg levels.

when I do the mikrotik bandwidth test, all appears great. I can get 10 megs or more. My pings time are 33ms to the 8th hop out.

2. 1 client uses Lingo VoIP and that day I upgraded his node, the VoIP stopped working completely. That have asked that I check to make sure the ports 5060 and 5065 are not blocked. I have double checked everything and I know I am not blocking those ports. I have not messed with the firewall settings in months.

3. I have a watchdog program that pings each node every minute of every day. Before the upgrade I would have 100% uptime of 95% of the network. Occasionally I would drop a packet here or there. But in gerneral it was perfect. Now I am missing 30 -50 packets each day on all nodes.

any ideas?

WHen did you do the upgrade to 2.9.18, and when did the problems begin?

[surfnet]

I upgraded last week, and the clients started calling within 24 hours. I now have a list of of 10 clients, all with great SNR and 11meg bitrate, all complaining. They are on many differnet nodes, and many different locations.

We had been trouble shooting each one, till we started to see a pattern emerge, and then we thought about the last thing we did, which was to upgrade all 22 nodes from 2.9.6 to 2.9.18.

I guess I could downgrade.. but I am the type of person that want to move forward, not backwards.. so I really want to fix this problem.

[uldis]

have you contacted the support? make the support output file when you see troubles on your APs and send it to support@mikrotik.com

[Equis]

Could you turn off all shaping/queue just for an hour or so to see if that affecting it?

[surfnet]

I certainly can do both of those suggestions. I will contact cleint and verify slow bandwidth, make support file, then remove queue and try again, and send that support file as well.