Hi, we´re using Cloud Core router with routeros v6.6. We have more than 2.000 simple queues with default queue type.
Every user has setted up MaxLimit for download for example 6M (depends on tarif) and 1/2 of download limit for upload.

When the user´s activity is very high (usually in the evening), download speed is going rapidly down.
For example: I have limit at 10M, so my speed should be 10Mbit for downloading, but every speedtest (speedtest.net) is only about 2 or 3Mbit.
If i dsabled my queue, my speed was unlimited (40-70Mbit), but when i setted up "unlimited" to the max limit and my queue was enabled, my speed was only 5 - 8Mbit (i think that´s why my queue was aplicated to me dogether with another active users)
In the morning when the user´s activity is not so big, i am able to reach 8-9Mbit.

BUT ! On routeros v5.x simple queues were ok, if i setted up for me 10M, i had 10Mbit and everything was working correctly.
Something is wrong on routeros v6.5 and higher with simple queue, for us ISP it is big problem.

Could anyone help please ? Thank you.

You will have to give more info, e.g. what is the structure, parents, children, priorities, etc.

It's without any parents and children, every SQ is separatly, priority 8 as default.

I have this problem (1700 dynamic simple queues) is added to a memory leak and crash

but the support says we do not see a problem

I have the exact same problem on 6.2 (CCR).

Simple Queues are not accurate. When 20mbit max is set the target can not push more than 12mbit.

I have the exact same problem on 6.2 (CCR).

Simple Queues are not accurate. When 20mbit max is set the target can not push more than 12mbit.

What queue type are you using? If using default-small, use default.
Or edit default-small to be 50 packets.

We were trying everything with queue type, default-small, default with queue size 50 and higher, nothing helped. I think it must be a bug in version 6.x with SQ.

I have this problem (1700 dynamic simple queues) is added to a memory leak and crash

but the support says we do not see a problem

Support emailed you 2 hours ago. The files didn't show any crashes, do you have UPS to ensure power stability? If it's not a power issue, you could send us remote access to monitor your device. I also see you use v6.4, since then, we have fixed some CCR crashes, so please upgrade to v6.6 first.

I also see in your graph, that traffic is going all the time. There is no indication of a crash.

About Speedtest - it's not the best way to test if your queues are working, because Speedtest only uses one TCP connection. This will not overload your link to the maximum.

Multiple TCP connection test (like torrent download or multiple FTP connection download) or UDP test will show the correct results. If your customers or users are complaining because of Speedtest, you could use the burst funcion - for some period of time (if customer hasn't used his traffic fully for a while) he wil be able to get bonus speed - speedtest will show more, but overall limit will remain the same.

I have this open ticket to 01/07/2013.
Starting with version 6.0, I have a memory leak, the support was responsible for months.
I'll even send video but I was given only one answer
Hello,

Sorry for delayed reply, unfortunately we still are not able to reproduce this
memory leak.
Try to set PPTP server max-mtu and max-mru to 1450. It is the new default value
for PPTP server, it was reduced to avoid fragmentation. Perhaps it can help.

Regards,
Janis Becs

I'm having this problem on 1100AHx2, x86.
Finally, I redid the configuration and removed the simple queues (now the tree queues).
But I still occasionally drops hardware, although no longer leaks memory with trees

We are seeing this problem as well - disable the simple queue for the client and it works fine. We have around 1100 simple queues.

I'm also having a similar problem, the customer has the limitation in queue, however he sails with much more speed, and shows no traffic queues, everything is reset in the counters on queue, it happens using hotspot
I'm using the CCR-1016 12G ROS v6.6

I'm also having a similar problem, the customer has the limitation in queue, however he sails with much more speed, and shows no traffic queues, everything is reset in the counters on queue, it happens using hotspot
I'm using the CCR-1016 12G ROS v6.6

You have not sorted the Queues by number (first column), so we can't see if your first queue has traffic.

then one of the things I did was insertthe queues below
/ ip hotspot user profile
SET [find default = yes] queue-insert-before = bottom

but I managed to work around the problem temporarily excluding the queue hs-<hotspot1> generated by the rate-limit(this blank) hotspot profile

I have the same problem with Simple Queues. Prior to version 6.1 did not pass me. I upgraded to 6.6 and 6.7 and even doing a downgrade to 6.1 I no longer work well, gives me less value and is not linear, it does so in a staggered manner.

Is there a solution?

I had similiar problem. I can't remember what i did, but i had something set to queue type ethernet-default kind=none and then my downloads sky rocketed . it was something to do with pfifo.

It was a 16Mbit customer struggling as soon as i changed the type they got it all

My 2 cents

Sew