Global Parent for Simple Queue really needed?
Hi admins,
that is my first attempt in EN section, lets how it goes
ROS v7.5
As I got my mikrotiks around two years ago, the vendor created a basic configuration which includes:
Thanks to a discussion with BRMateus2 about PCQ Queue size and total queue size recommendations for 1G/50M internet I upgraded the simple queue configuration but still not using a global parent as it was confirmed that ROS will handle ensure that bandwidth is equally distributed.
and
BUT recently I noticed, if a VLAN continuously downloads it consumes all bandwidth although minimum bandwidth is set for User 3.Minimum bandwidth is ignored but why?
but no overload on of the hardware:
so I double checked the configuration:
basic:
advanced:
So the only "workaround" I figured out, is to use a global entry
and making all previous top level entries a child of that one.
It only works if the target is set to 0.0.0.0/0, putting it on the uplink only does not have any affect.
My theory is, that is not known to the algorithm to which route/interface the 20M refer to, means it does not know that all VLAN interfaces eat from the same cake.So if User 6 continuously downloads it always eats free bandwidth if a different user makes a break ending up eating the full bandwidth.How to tell the algorithm that the 20M have to be shared between all top levels queues?
As my vendor uses the same approach as in the code block above for the initial configuration, I'm quite surprised that I need this global parent.
Any clue why?
thx
stefan
that is my first attempt in EN section, lets how it goes
ROS v7.5
As I got my mikrotiks around two years ago, the vendor created a basic configuration which includes:
/queue simple
add limit-at=100M/100M max-limit=1G/1G name=admin target=vlan20-admin.sfp+
add limit-at=100M/100M max-limit=1G/1G name=mieter1 target=vlan21-mieter1.sfp+
add limit-at=100M/100M max-limit=1G/1G name=mieter2 target=vlan22-mieter2.sfp+
add limit-at=100M/100M max-limit=1G/1G name=mieter3 target=vlan23-mieter3.sfp+
add limit-at=50M/50M max-limit=100M/100M name=wlan-gast target=br.wlan-gast
Thanks to a discussion with BRMateus2 about PCQ Queue size and total queue size recommendations for 1G/50M internet I upgraded the simple queue configuration but still not using a global parent as it was confirmed that ROS will handle ensure that bandwidth is equally distributed.
/queue simple
add limit-at=0/0 max-limit=0/0 name=Intranet_unlimited target=10.10.0.0/18 dst=10.10.0.0/18 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=USER1 target=VLAN_USER1 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=USER2 target=VLAN_USER2 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=USER3 target=VLAN_USER3 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=USER4 target=VLAN_USER4 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=USER5 target=VLAN_USER5 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=USER6 target=VLAN_USER6 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=USER7 target=VLAN_USER7 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=USER8 target=VLAN_USER8 queue=pcq-up/pcq-down
add limit-at=10M/10M max-limit=90M/90M name=GUEST target=VLAN_GUEST queue=pcq-up/pcq-down
add max-limit=10M/10M name=SERVICES-GLOBAL target=Combo1
add limit-at=1M/1M max-limit=10M/10M name=COMMON-SERVICES target=VLAN_COMMON queue=pcq-up/pcq-down parent="SERVICES-GLOBAL"
add limit-at=1M/1M max-limit=10M/10M name=SYSTEM-SERVICES target=VLAN_SYS queue=pcq-up/pcq-down parent="SERVICES-GLOBAL"
add limit-at=1M/1M max-limit=10M/10M name=IoT target=VLAN_IoT queue=pcq-up/pcq-down parent="SERVICES-GLOBAL"
[adminuser@CR_2B_MT-CCR1009-8G-1S-1S+] /queue/type> print
Flags: * - default
5 name="pcq-up" kind=pcq pcq-rate=0 pcq-limit=4KiB pcq-classifier=src-address,dst-address pcq-total-limit=400KiB pcq-burst-rate=0 pcq-burst-threshold=0 pcq-burst-time=10s pcq-src-address-mask=32 pcq-dst-address-mask=32 pcq-src-address6-mask=128
pcq-dst-address6-mask=128
6 name="pcq-down" kind=pcq pcq-rate=0 pcq-limit=4KiB pcq-classifier=src-address,dst-address pcq-total-limit=400KiB pcq-burst-rate=0 pcq-burst-threshold=0 pcq-burst-time=10s pcq-src-address-mask=32 pcq-dst-address-mask=32
pcq-src-address6-mask=128 pcq-dst-address6-mask=128
BUT recently I noticed, if a VLAN continuously downloads it consumes all bandwidth although minimum bandwidth is set for User 3.Minimum bandwidth is ignored but why?
but no overload on of the hardware:
so I double checked the configuration:
basic:
advanced:
So the only "workaround" I figured out, is to use a global entry
1 name="UPLINK" target="" parent=none packet-marks="" priority=8/8 queue=default-small/default-small limit-at=0/0 max-limit=20M/20M burst-limit=0/0 burst-threshold=0/0 burst-time=0s/0s bucket-size=0.1/0.1
It only works if the target is set to 0.0.0.0/0, putting it on the uplink only does not have any affect.
My theory is, that is not known to the algorithm to which route/interface the 20M refer to, means it does not know that all VLAN interfaces eat from the same cake.So if User 6 continuously downloads it always eats free bandwidth if a different user makes a break ending up eating the full bandwidth.How to tell the algorithm that the 20M have to be shared between all top levels queues?
As my vendor uses the same approach as in the code block above for the initial configuration, I'm quite surprised that I need this global parent.
Any clue why?
thx
stefan
Bitte markiere auch die Kommentare, die zur Lösung des Beitrags beigetragen haben
Content-ID: 4815833729
Url: https://administrator.de/contentid/4815833729
Ausgedruckt am: 21.11.2024 um 17:11 Uhr