*) added Combo-Mode setting for combo ports;
*) allow to configure tx & rx flow control separately (only tx flow control
is enabled by default) and make them work;
*) report current status of rx/tx flow control;
*) report correctly rx/tx pause frames under Errors tab;
*) report that link is paused under Links tab;
*) report tx queue size under Stats tab;
*) make sure that SwOS is still responsive when most of SFP modules are plugged in;
*) fixed problem where switch locked up after a while and only small packets could be sent;
*) fixed problem where sometimes by loading bigger backup files switch
went into reboot loop;
*) make LACP keep working even if only one port is active in trunk
(helps with fail-over);
*) fixed problem where changing LAG mode from static to active did not work;
*) CRS328C & CRS326S: make switch accessible by web
if all SFP modules are plugged in;
*) CRS312: fixed combo ports LEDs;
*) CRS312: fixed problem where traffic on combo ports stopped working after reboot;
*) CRS328C: added support for 100Base-FX modules;
*) CRS328C: make DAC link work with CCR boards regular SFP ports;
*) CRS326Q: allow to force 10Gx4 or 40G mode on QSFP ports;
*) CRS326Q: report QSFP module status;
*) CRS328P: report correctly PoE values via SNMP;
*) CRS309: added support for ETH/BOOT port configuration;
*) CSS106: report correctly PoE & SFP values via SNMP;
*) CSS106: fixed IGMP packet forwarding when snooping is enabled;
*) CSS106: fixed SFP LED;
*) CSS106: fixed ACL, sometimes it did not work if MAC & IP address matchers
were used at same time (or when Allow-From address was specified);
*) CSS106: handle properly unknown unicast packets (do not try to send them to CPU),
otherwise under load it impacts forwarding speed of the switch;
*) CSS106: do not allow to access switch by VLAN from port that is not member
of that VLAN if port is in strict or enabled VLAN mode;
*) CSS106: make static hosts work in IVL mode;
Hi Becs, reading the improvements, I am wondering if some of the issues I have seen on a CAPAC connected to my 260GS is related to or more accurately could cause mac address issues on the switch or back and forth with the main router. Thinking it was the capac I replaced it with a tplink eap245, but its also seeing same locked up issues.
Currently connected direct from CCR1009 to 260GS comboSFP port to SFP port respectively. I had the 260GS connected via a 24 port switch from a hex previously with same wifi issues.
Items that may be related
small packets locking up switch temporarily
-handling of small packets
-vlan port leakage
So Far two strikes and cannot recommend attempting the udpate on the 260GS models yet.
First 260GS - will not take upload tried at least 5 times, once using automated method other times via manual upload
Second 260GS - tried manual first time, it seemed to take as it showed Current as 2.12 but could not enter in any menu items and thus logged out and was no longer visible on winbox or accessible via direct lanip entry. Suspect that the config was somehow changed by the upload then??
First 260GS does not want to update.
This is par for the course, everytime I update these units its like surgery with an infection.
Does anybody at MT actually use these units and test them?
Okay I can connect to the switch via the homelan on one of its untagged ports and it seems to be fine.
Not sure why it cannot be accessed ‘over the network’ via its trunk port??
Just to confirm there is no downside to accepting default settings
On all port flow control is enabled and auto negotiation is enabled??
On RSTP all ports have RSTP enabled (with Bridge priority at 8000 and port cost set to short)??
Special Item - I noticed WATCHDOG is checked/enabled. What does this do as I see no options of other settings for this which would be expected right??
(mikrotik discovery protocol is also enabled)
Very weird, I can log into and acces the 260GS from all ethernet ports.
I have a cable coming from my old router netgear being used solely as an ap/switch on VLAN11
POrt 1 is untagged vlan11
Port 2 is untagged vla
Port 3 is trunk (includes vlan11)’
POrt 4 is untagged vlan55
Port 5 is untagged vlan59
SFP is Trunk from router includes vlan11
Why am I able to login to the switch from all ports on the 260gs - is it because the the gigabit dumb/switch is allowing this??
Okay I re-deployed the switch in the basement and now its fully accessible and all the devices operating normally from the Trunk SFP port.
I agree mozerd, I think I have haunted switches LOL Still trying to upload the first switch with 2.12.
By the way I dont dare tough any of the security access to switch settings.
I have all the ports able to access the switch and didnt narrow it down by vlan or anything as everytime I try do to that i get locked out no matter what and have to start from defaults (I believe also goes back to first firmware on the device too…)
Finally got the first one to upgrade, I kept hitting auto-download install on the 4th try it worked.
Same issue after upgrade, its no longer visible to winbox and not accessible through direct IP entry.
Looks like I will have to take it out and connect it up via my PC to magically let it be visible again, this is so frustrating.
Yup connected it directly at another location and now its working again back in situ… most weird,
Two major improvements for me. First is the intermittent lockup issue that has plagued SwitchOS for quite a few releases. I will be repeating the test that I did back in April that caused both switches to lock up almost instantly. The other major improvement for me is that LACP will now work properly with only one link. Kinda defeated the purpose of LACP as a redundancy tool if the whole LAG failed if one of the links failed. I will be testing that as well.
Updated one CSS106-5G-1S, one CRS326-24G-2S+, and two CSS326-24G-2S+ without issue. The CSS106 was the only one that showed that 1.12 was available - the other three I had to do a manual upload. The CRS326 took quite a bit longer. For each update I had a continuous ping to some device that required passing packets through the switch being updated. For the three CSS switches, I dropped one or two pings, but for the CRS326, I dropped six pings before traffic resumed being passed.
Obviously I did not attempt to update my RB260GS (old hardware) since it is not compatible with 2.x firmware
I have now performed both my LACP test and enabled IPv6 which caused both switches to crash earlier this year. Here is a simplified drawing of my LAN.
Last April, I had enabled IPv6 on router #1 and essentially instantly, both switches crashed. This morning after installing 2.12, I enabled IPv6 on router #1 without any issues with the switches. I have a very small sample size - one incident - where both switches crashed after enabling IPv6. Was that the cause? I don’t know. All I know for sure is that enabling IPv6 on router 1 this morning did not cause issues with either switch.
For the LACP test, I have to explain a little. The VLAN trunk between the two switches is currently a 1G copper connection. As you can tell from my drawing, I am exceedingly dependent on that link. If the link between the two switches fails, I loose essentially everything. One of these days, I will get a second link between the switches so that I have a redundant route - I need to install some conduit first. I have Mikrotic S85DLC05D multimode fiber SFPs in both switches, and both ends are configured exactly the same as the copper links. I have the two ports configured as Active on the LAG page on the family room switch and passive on the garage switch. For the test, I temp ran a 75 foot fiber jumper between the two switches. Before plugging in the FO jumper at the family room end, I initiated several continuous pings to various devices. When I plugged in the FO jumper, the LAG pages on both ends showed that the LAG was active, and no pings dropped. Looking at the stats page confirmed that traffic was being carried on both links. I then disconnected the copper link, and after one ping drop, all traffic was now going over the fiber link. Plugging the copper link back in, and again the LAG and stats pages showed that traffic was routing over both links. Repeated the test several times back and forth. The only thing I noticed was that when disconnecting the FO link, there was a longer ping drop before the all the traffic switched to the copper link - about 15 seconds. BTW, I had previously proved in the past that a LAG with only one working member would fail.
Initial testing shows a substantial improvement from 2.11 in these two areas.
@anav - if the upgrade fails on CSS106, toggling the RSTP setting on ports may help. As a last resort follow this article - Cannot upgrade SwOS. Regarding other issues you are facing, please report them separately on support portal or email to support@mikrotik.com.
Still 1 hour after upgrade from 2.11 to 2.12 there are pause frames between 10GB SFP mkt DAC from CSS326-24G-2S+(RX pauses) to CRS328-24P-4S+(TX pauses), however its only one direction instead of both direction without 1 direction doubling as in the past…
Thanks EdPa, All good now. THey are on static IPs set on the switch and the router so I think I just had to somehow kickstart them to be seen over the trunk ports. Weird but true. As far as the upgrade why they take multiple attempt to upgrade is beyond my comprehension. They are both good now.
If I had been stuck would have done the portal route.
