NOTE: in this build we have applied a patch that should solve most of the queueing related performance issues on CCR.
If these changes will provide positive feedback we will consider to put them into release of v6.19.
Please, test queues on CCR and report results and any problems you might find.
The changes also include long term stability imrovements
What’s new in 6.19rc6 (2014-Aug-08 12:38):
*) CCR - major queueing performance fix
*) CCR - fixed crash on L2MTU change;
*) CCR - improved performance;
*) firewall - packet defragmenting will only happen with connection tracking enabled;
*) firewall - optimized option matching order with-in a rule;
*) firewall - rules that require CONNTRACK to work will now have Invalid flag when CONNTRACK is disabled;
*) firewall - rules that require use-ip-firewall to work will now have invalid flag when use-ip-firewall is disabled;
*) firewall - rules that have interface with “Slave” flag specified as in-/out-interface will now have Invalid flag;
*) firewall - rules that have interface without “Slave” flag specified as in-/out-bridge-port will now have Invalid flag;
*) firewall - rules with Invalid flags will now be auto-commented to explain why;
*) console - added unary operator ‘any’ that evaluates to true if argument
is not null or nothing value;
*) l2tp - force l2tp to not use MPPE encryption if IPsec is used;
*) sstp - force sstp to not use MPPE encryption (it already has TLS one);
*) sstp - report TLS encryption as well;
I have seen the exact same issues concerning the CCR using mangles and simple queues after version ~6.14, the most frustrating thus far is the Queue/Mangle Performance but another feature that also has not been mentioned is that all SNMP services running on our CCR’s run for a few hours after reboot and after that an snmpwalk command shows ZERO data. I have tried start/stopping the service with no success the only thing that works is a reboot.
I see that this issue has not been addressed yet in the v6.19 changelog?
Mail was delayed so i write result here
It did not improve performance at all for my configuration and make router unresponsive.
Actually performance was lower and system locked during speed test.
supout sent and waits for retry in mail queue…
Installed 6.19rc6 on my CCR1009. I have dnat rules for mail and www. dnat rules involving port 443 never reaches the target host and just timeout. Going back to 6.18 solve this issue. Other ports work fine.
On CCR1016-12S-1S+ I set SFP1 & SFP2 to 9000 L2MTU, added a 802.3ad bonding interface, with interfaces SFP1 & SFP2 as slaves, then added a VLAN interface to bonding1.
The result is the VLAN interface (and bonding 1) will not pass traffic, and SFP10 / SFP11 / SFP12 interfaces light up / blink like there is traffic with nothing connected to them.
traffic about 300Mbit/s load 30%, active users about 1000
if traffic > 300Mbit/s loop reboot every 1 min load 100%
RoS6.18 500-600Mbit load 70%, works stability. But in BHCA we have about 1.5 Gbit/s, so we didn’t tested MT in this time.
RoS6.19rc6
80 magle rules for mark traffic 20 HTB queues in tree pcq
700 Mbit/s avg 50% ( but when we switch MT for Linux, where is 1.2Gbit/s)
RoS6.18 400Mbit/s avg 60% (but when we switch MT for Linux, where is 1.2Gbit/s)
so "queue tree and simple " have a sad perfomance.
Linux+tc+ipset+iptables = 1.5G avg 50% in same configuration ( 8500 queues) in BHCA
The biggest improvements are in large numbers of simple queues, not queue tree setup. But please send a supout.rif file and a description of your setup to mikrotik support, we will see how to improve your situation.
After the 6:17 version only problems.
CCR and RB1100 presenting as PPPoE server reboots. RBs three new 433AH we put into operation last week with version 4:17 reboot every 10 or 15 minutes and had intakes of 100% CPU at times.
