Ether4 fcs error on link

Fcs error on link mikrotik что это

Sun Dec 29, 2019 10:32 pm

My network completely stopped. Couldn’t get anything to work. Couldn’t statically set an IP and reach the router.

Rebooted router. nope.
Noticed even my caps had stopped transmitting.

Rebooted switch. Everything came back.

Looked at the log from my router.
Sfp-sfpplus1 fcs error on link

After rebooting the switch.
Sfp-sfpplus1 link down
Sfp-sfpplus1 link up (Speed 10G, full duplex)

Few seconds later. Everything got IP addresses and returned to normal.

Re: FCS error on link

Sun Dec 29, 2019 10:42 pm

Re: FCS error on link

Sun Dec 29, 2019 11:39 pm

Unit had been in operation for a few weeks at this point.

During lockup.
It still provided POE to a hAP AC2 and to a 5 port 260.

Re: FCS error on link

Mon Dec 30, 2019 9:38 am

Re: FCS error on link

Mon Dec 30, 2019 6:02 pm

Re: FCS error on link

Mon Dec 30, 2019 7:21 pm

Re: FCS error on link

Mon Dec 30, 2019 7:22 pm

Re: FCS error on link

Mon Dec 30, 2019 8:16 pm

Re: FCS error on link

Mon Dec 30, 2019 11:10 pm

The FCS error was what the router was reporting when the switch was locked up.

That cable had been used before with another manufacture’s switch, and the switch never locked up.

The cable is a direct connect type where the modules that go into the router and switch are part of the cable.

I was hoping I was providing meaningful data that could be used to CORRECT whatever made this happen.

And remember. This is the first time I observed this problem at the office, and the system had been running for over a week. Infact. It was working one second then not. I was watching something on Amazon Prime on my computer at the time. I only noticed when the error screen popped up.

Источник

Fcs error on link mikrotik что это

ether2 fcs error on link

У меня такая же ошибка уже месяца полтора-два.
Делал то что и Вы, менял провода, порты, настройки.
Ждал новую прошивку, в ней было описание что вроде это пофиксили,
но увы, снова и снова. ошибки есть.

Кстати, думаю что это связано с насхлёстом одного уровня над другим.
У меня получается сделан бондинг (ether1 и ether2) , на этом бондинге подняты виланы (много) и уже на виланах идёт адресация. Также туннели висят на некоторых адресах.
Думаю что из-за вот этих 2х уровней абстракции + туннели и возникает. проблемы.

В другой конторе почти на 80% такая же конфигурация, но там ошибок нет. Так что думаю или дело в туннелях (там их нет, а бондинг и виланы есть),
или всё же с другой стороны микротика, на свиче что-то с портами Ethernet (мож свитч подыхает?)

Опишите Вашу конфигурацию. может найдём общие точки соприкосновения!?

Пару раз сталкивался с данной проблемой.
В первом случае заменил инжектор во втором случае больно пинал монтажника который кабель положил вдоль силовой линии да еще и обжал криво

Вобще года 3 тому назад я вложил несколько десятков тысяч денег в подбор кабеля. Перепробовал массу производителей от китайского овна до мега дорогих. Сейчас юзаю кабель для внешней прокладки от нетлан, еще ни разу не подвел. Соотношение цена/качество отличное

Источник

Fcs error on link mikrotik что это

Sun Oct 04, 2015 4:12 pm

Yesterday replaced a rb1100AH running v.6.30.2 to a brand new CCR1009-8G-1S-1S+ router.

Looked in the log today and found it filled with fcs errors?

All ethernet cables are almost new and connect to new routers and on the rb1100AH showed no errors and were Gigabit connected (auto negotiation on)

Now these failures?
I installed the CCR with 6.32.2 but reverted back to 6.30.2 as the most stable version but the problem didn’t disappear.

1. It can’t be the cables, they were all fine on the rb1100AH and basically all new.
2. It can’t be the software? 6.32.2 or 6.30.4 makes no difference. And most of my other routers now work fine with either one of these two versions.

I ran 2 ping tests with 100ms timeout to adjacent router and that is fine. 1st is 100ms and 50 bytes packet size, 2nd is 100ms but 1500 packet size. On the latter I see an occasional time out and I see 1% package loss.
The stats of the interface is also showing the error. 2100 errors in 25 mins uptime.

Do I have a failing CCR? Where is this error coming from?

Show your appreciation of this post by giving me Karma! Thanks.

WISP operator based on MT routerboard & ROS.

Re: fcs error on new CCR1009

Tue Oct 06, 2015 11:19 am

Show your appreciation of this post by giving me Karma! Thanks.

WISP operator based on MT routerboard & ROS.

Re: fcs error on new CCR1009

Tue Oct 06, 2015 7:37 pm

Three things cause FCS errors typically, with the first two being the most likely.

1) Physical cabling issues

2) Speed / duplex mismatch or errors

Steps to resolve.

1) Change the cable with a known good one on the link you are getting FCS and reset counters.

2) Hard code speed and duplex to match on both sides of the link.

3) Move to the bugfix version — 6.30.4 as it is the most stable version for ROS right now in general

Re: fcs error on new CCR1009

Tue Oct 06, 2015 7:50 pm

be sure you have updated the firmware (not routeros)

put a manageable switch as a bypass on that link to see if switch receives the same errors

Re: fcs error on new CCR1009

Tue Oct 06, 2015 8:14 pm

Three things cause FCS errors typically, with the first two being the most likely.

1) Physical cabling issues

2) Speed / duplex mismatch or errors

Steps to resolve.

1) Change the cable with a known good one on the link you are getting FCS and reset counters.

2) Hard code speed and duplex to match on both sides of the link.

3) Move to the bugfix version — 6.30.4 as it is the most stable version for ROS right now in general

Show your appreciation of this post by giving me Karma! Thanks.

WISP operator based on MT routerboard & ROS.

Re: fcs error on new CCR1009

Wed Oct 07, 2015 5:35 pm

Really sounds like you may have a bad board out of the box. few comments below

1) If you can swap the 1100 back in and the FCS goes away, then i would agree cabling is prob not it

2) This is true only on copper (and you can set 100M Full on each side just for testing) SFP ports can be hard coded to 1000/Full as well.

3) In our lab we have a CCR1009 connected to MikroTik, Cisco and HP network equipment on copper and fiber without issue. unfortunately you probably got a bad CCR.

There was a batch of bad CCR1009s early on and it’s possible you got one that’s been on the shelf for a while as all the new CCR1009s shipping seem to be pretty solid.

Re: fcs error on new CCR1009

Wed Oct 07, 2015 6:49 pm

Re: fcs error on new CCR1009

Thu Oct 08, 2015 11:53 am

Nailed the issue for 95%:

All links / ethernet cables leave from CCR1009 over some 15 meter of ftp cable (all well grounded on both ends. ) to a tower cabinet.
In the cabinet a 24V PoE battery fed midspan power inserter with lighting/surge protector is fit. From here some 8-10 meter of ftp (all well grounded on both ends. ) run up the tower to connect to the radios. Mostly Netmetals.

The midspan PoE injector is from Cyberteam, Poland. This is a managed Gigabit version. This is actually the 3rd box that give me problems. One just died, the second had no more webserver working and became useless because of that en this last one after 2 months again fell without web server control. So my suspicion went to this device. Last night I replaced it to a brand new unmanaged gigabit net-protector and almost all fcs error disappeared from the logs.
So this is the 3rd managed netprotector heading for the bin (with 2 unmanaged going the same before. poor quality stuff. )

I still suspect the last remaining errors, now only on one link, are due to this new netprotector. But to be honest there are not a lot 24V multiple port passive PoE midspan injectors in gigabit version with IP management on the market. The nearest solution I can find are Netonix WISP switch but I need to convert it into 1:1 port to port switch only since the several routes have different dhcp-servers and hence these ‘pipe’s need to stay separated.

To reply to some of your comments;

3) In our lab we have a CCR1009 connected to MikroTik, Cisco and HP network equipment on copper and fiber without issue. unfortunately you probably got a bad CCR.

There was a batch of bad CCR1009s early on and it’s possible you got one that’s been on the shelf for a while as all the new CCR1009s shipping seem to be pretty solid.

Show your appreciation of this post by giving me Karma! Thanks.

WISP operator based on MT routerboard & ROS.

Re: fcs error on new CCR1009

Thu Oct 08, 2015 11:56 am

Show your appreciation of this post by giving me Karma! Thanks.

WISP operator based on MT routerboard & ROS.

Re: fcs error on new CCR1009

Thu Oct 08, 2015 3:31 pm

Last night at around 1 am we replaced the gigabit netprotector (Poe midspan injector) for another more simple one (Also giga, also Cyberteam, but no remote access, no control whatsoever) brand new out of the box.
Immediately after the change the fcs errors didn’t come back. Well, instead of every 30 seconds until this morning 10 am when I logged in again the log only showed 4 fcs errors in total since the change. good!
. You would think. =>

While looking over a remote winbox session in this CCR1009’s log this morning I saw the errors coming back! First every 10 mins, then every 5 mins, every 3-4 mins and now they are back to every 2 mins.
How is this possible? Overnight the problem vanished just to come back in the morning. Very, very weird.

I don’t know what to do now. Cannot go to 100mbps setting on each end since this is my main backhaul that at times has more traffic than 100mbps. Fast Ethernet will create a bottleneck again. (I spend a lot of time and money and energy into reforming a previously duo bonded link with double antenna sets etc. high in towers into a ‘ac’ single 40Mhz wide link to get up to 300Mbps over this new link. (tested up to 280M!)

I also cannot fit in the gigabit power inserters from Mikrotik. Because the poe insert is in the middle of a cable. Both ends need to be at least 10 meters.
I don’t seem to be able to find any passive poe midspan inserters that work on 24V battery power and have gigabit port and are remotely controllable.

So, main question; Until I find any solution, how bad is it these errors are there? They fill the log, but do they have any other harm? I don’t seem to notice problems (yet?). Ping on long and short package see no losses and ping times are good and traffic flows.
But out of the 12 NetMetals installed over the last weeks, all running with latest ROS and firmware in bridge mode (fastpath enabled) I have 3 of them that just stop passing traffic once a 2 or 3 weeks. only a power reboot brings these back in working order again. Can this have anything to do with some fcs error too? (Since the Netmetals don’t show these errors in the logs I never looked for them. )

Show your appreciation of this post by giving me Karma! Thanks.

WISP operator based on MT routerboard & ROS.

Источник

Fcs error on link mikrotik что это

Mon Apr 10, 2017 12:28 pm

Yesterday (the same as before about 3/4 year) to WAN IFC CCR appeared on «FCS Error on interface» error. At that time we thought it was a fault mikrotik and did not pay it much attention.

IFC could turn and errors disappeared.
But this morning it appeared again and turning the IFC, it decides not solved During the 15 minutes it was so 1260 RX FCS Error statistics IFC.
Thus, unlike the day before yesterday against a «Back then» was that the IFC normally ping and communication over IP function (the question of how long it would take).

What we tried:
1) gradually turn off IP addresses in an attempt to isolate the communications that fall — without success, they doing it all
4) disable IFC and after a while switch on — no Results

Finally, in this case helped disable all addresses in Adress list and gradually one by one turn on, turn off, then turn the other one on and off, to turn each of them has always been for a while primary address, at that moment dana address is always started from the outside Ping this style we all eventually turned and began to work, but not a solution.

Yesterday, on the recommendation of the man in our housing turn off Auto negotiation on an Ethernet port, and set to fix 100Mbit full duplex.
When I turned on him again today (auto negotiation) and FCS Error disappeared (according to IFC well log set 100Mbit Fullduplex)
So I think it’s some kind error somewhere .
Technic guy said he had also solved it with Mikrotik and that was just a problem in that the IFC runs at 100Mbit — which is the only difference from previos housing (there was 1Gbit O2 is 100 Mbps per line)
Netwatch meanwhile is set to GW and if there is a timeout 5s (once per minute), and it turns IFC WAN (it solves unavailability), but does not generally FCS Error.

So if it happens again, there will be an interesting test by me throw it on the RB2011 or whatever it is, because he has a physical 100Mbit interfaces. Optionally upgrade from 100Mbit to 1Gbit (which is unnecessary, peak traffic was 35Mbit)
I must say that on the other (admin router — the RB2011) it has not happened yet not once — there is a 100Mbit link, so maybe it’s best estimate

What do you think?

log:
.
xxxx interface,warning ether1 — WAN fcs
error on link
xxx interface,warning ether1 —
WAN fcs error on linkxxxx interface,warning
ether1 — WAN fcs error on linkxxxx
interface,warning ether1 — WAN fcs error on link
xxxx
xxx interface,warning ether1 — WAN fcs error on linkxxxx interface,warning ether1 — WAN fcs error
on link
.

Источник

Ethernet

Sub-menu: /interface ethernet
Standards: IEEE 802.3

MikroTik RouterOS supports various types of Ethernet interfaces — ranging from 10Mbps to 10Gbps Ethernet over copper twisted pair, 1Gbps and 10Gbps SFP/SFP+ interfaces and 40Gbps QSFP interface. Certain RouterBoard devices are equipped with a combo interface that simultaneously contains two interface types (e.g. 1Gbps Ethernet over twisted pair and SFP interface) allowing to select the most suitable option or creating a physical link failover. Through RouterOS, it is possible to control different Ethernet related properties like link speed, auto-negotiation, duplex mode, etc, monitor a transceiver diagnostic information and see a wide range of Ethernet related statistics.

Properties

Property	Description
advertise (10M-full | 10M-half | 100M-full | 100M-half | 1000M-full | 1000M-half | 2500M-full | 5000M-full | 10000M-full; Default: )	Advertised speed and duplex modes for Ethernet interfaces over twisted pair, only applies when auto-negotiation is enabled. Advertising higher speeds than the actual interface supported speed will have no effect, multiple options are allowed.
arp (disabled | enabled | local-proxy-arp | proxy-arp | reply-only; Default: enabled)	Address Resolution Protocol mode: disabled — the interface will not use ARP enabled — the interface will use ARP local-proxy-arp — the router performs proxy ARP on the interface and sends replies to the same interface proxy-arp — the router performs proxy ARP on the interface and sends replies to other interfaces reply-only — the interface will only reply to requests originated from matching IP address/MAC address combinations which are entered as static entries in the ARP table. No dynamic entries will be automatically stored in the ARP table. Therefore for communications to be successful, a valid static entry must already exist.
auto-negotiation (yes | no; Default: yes)	When enabled, the interface «advertises» its maximum capabilities to achieve the best connection possible. Note1: Auto-negotiation should not be disabled on one end only, otherwise Ethernet Interfaces may not work properly. Note2: Gigabit Ethernet and NBASE-T Ethernet links cannot work with auto-negotiation disabled.
bandwidth (integer/integer; Default: unlimited/unlimited)	Sets max rx/tx bandwidth in kbps that will be handled by an interface. TX limit is supported on all Atheros switch-chip ports. RX limit is supported only on Atheros8327/QCA8337 switch-chip ports.
cable-setting (default | short | standard; Default: default)	Changes the cable length setting (only applicable to NS DP83815/6 cards)
combo-mode (auto | copper | sfp; Default: auto)	When auto mode is selected, the port that was first connected will establish the link. In case this link fails, the other port will try to establish a new link. If both ports are connected at the same time (e.g. after reboot), the priority will be the SFP/SFP+ port. When sfp mode is selected, the interface will only work through SFP/SFP+ cage. When copper mode is selected, the interface will only work through RJ45 Ethernet port.
comment (string; Default: )	Descriptive name of an item
disable-running-check (yes | no; Default: yes)	Disable running check. If this value is set to ‘no’, the router automatically detects whether the NIC is connected with a device in the network or not. Default value is ‘yes’ because older NICs do not support it. (only applicable to x86)
tx-flow-control (on | off | auto; Default: off)	When set to on, the port will generate pause frames to the upstream device to temporarily stop the packet transmission. Pause frames are only generated when some routers output interface is congested and packets cannot be transmitted anymore. auto is the same as on except when auto-negotiation=yes flow control status is resolved by taking into account what other end advertises.
rx-flow-control (on | off | auto; Default: off)	When set to on, the port will process received pause frames and suspend transmission if required. auto is the same as on except when auto-negotiation=yes flow control status is resolved by taking into account what other end advertises.
full-duplex (yes | no; Default: yes)	Defines whether the transmission of data appears in two directions simultaneously, only applies when auto-negotiation is disabled.
l2mtu (integer [0..65536]; Default: )	Layer2 Maximum transmission unit. Read more>>
mac-address (MAC; Default: )	Media Access Control number of an interface.
master-port (name; Default: none)	Outdated property, more details about this property can be found in the Master-port page.
mdix-enable (yes | no; Default: yes)	Whether the MDI/X auto cross over cable correction feature is enabled for the port (Hardware specific, e.g. ether1 on RB500 can be set to yes/no. Fixed to ‘yes’ on other hardware.)
mtu (integer [0..65536]; Default: 1500)	Layer3 Maximum transmission unit
name (string; Default: )	Name of an interface
orig-mac-address (read-only: MAC; Default: )	Original Media Access Control number of an interface.
poe-out (auto-on | forced-on | off; Default: off)	Poe Out settings. Read more >>
poe-priority (integer [0..99]; Default: )	Poe Out settings. Read more >>
sfp-shutdown-temperature (integer; Default: 95)	Temperature at which interface will be temporarily turned off due to too high detected SFP module temperature (introduced v6.48).
speed (10Mbps | 10Gbps | 100Mbps | 1Gbps; Default: )	Sets interface data transmission speed which takes effect only when auto-negotiation is disabled.

Read-only properties

Property	Description
running (yes | no)	Whether interface is running. Note that some interface does not have running check and they are always reported as «running»
slave (yes | no)	Whether interface is configured as a slave of another interface (for example Bonding)
switch (integer)	ID to which switch chip interface belongs to.

Menu specific commands

Property	Description
blink ([id, name])	Blink Ethernet leds
monitor ([id, name])	Monitor ethernet status. Read more>>
reset-counters ([id, name])	Reset stats counters. Read more>>
reset-mac-address ([id, name])	Reset MAC address to manufacturers default.
cable-test (string)	Shows detected problems with cable pairs. Read More >>

Monitor

To print out a current link rate, duplex mode, and other Ethernet related properties or to see detailed diagnostics information for transceivers, use /interface ethernet monitor command. The provided information can differ for different interface types (e.g. Ethernet over twisted pair or SFP interface) or for different transceivers (e.g. SFP and QSFP).

Properties

Property	Description
advertising (10M-full | 10M-half | 100M-full | 100M-half | 1000M-full | 1000M-half | 2500M-full | 5000M-full | 10000M-full)	Advertised speeds and duplex modes for Ethernet interfaces over twisted pair, only applies when auto-negotiation is enabled
auto-negotiation (disabled | done | failed | incomplete)	Current auto negotiation status: disabled — negotiation disabled done — negotiation completed failed — negotiation failed incomplete — negotiation not completed yet
default-cable-settings (short | standard)	Default cable length setting (only applicable to NS DP83815/6 cards) short — support short cables standard — support standard cables
full-duplex (yes | no)	Whether transmission of data occurs in two directions simultaneously
link-partner-advertising (10M-full | 10M-half | 100M-full | 100M-half | 1000M-full | 1000M-half | 2500M-full | 5000M-full | 10000M-full)	Link partner advertised speeds and duplex modes for Ethernet interfaces over twisted pair, only applies when auto-negotiation is enabled
rate (10Mbps | 100Mbps | 1Gbps | 2.5Gbps | 5Gbps | 10Gbps | 40Gbps |)	Actual data rate of the connection.
status (link-ok | no-link | unknown)	Current link status of an interface link-ok — the card is connected to the network no-link — the card is not connected to the network unknown — the connection is not recognized (if the card does not report connection status)
tx-flow-control (yes | no)	Whether TX flow control is used
rx-flow-control (yes | no)	Whether RX flow control is used
combo-state (copper | sfp)	Used combo-mode for combo interfaces
sfp-module-present (yes | no)	Whether a transceiver is in cage
sfp-rx-lose (yes | no)	Whether a receiver signal is lost
sfp-tx-fault (yes | no)	Whether a transceiver transmitter is in fault state
sfp-type (SFP-or-SFP+ | DWDM-SFP | QSFP+)	Used transceiver type
sfp-connector-type (SC | LC | optical-pigtail | copper-pigtail | multifiber-parallel-optic-1×12 | no-separable-connector | RJ45)	Used transceiver connector type
sfp-link-length-9um (m)	Transceiver supported link length for single mode 9/125um fiber
sfp-link-length-50um (m)	Transceiver supported link length for multi mode 50/125um fiber (OM2)
sfp-link-length-62um (m)	Transceiver supported link length for multi mode 62.5/125um fiber (OM1)
sfp-link-length-copper (m)	Supported link length of copper transceiver
sfp-vendor-name (string)	Transceiver manufacturer
sfp-vendor-part-number (string)	Transceiver part number
sfp-vendor-revision (string)	Transceiver revision number
sfp-vendor-serial (string)	Transceiver serial number
sfp-manufacturing-date (date)	Transceiver manufacturing date
sfp-wavelength (nm)	Transceiver transmitter optical signal wavelength
sfp-temperature (C)	Transceiver temperature
sfp-supply-voltage (V)	Transceiver supply voltage
sfp-tx-bias-current (mA)	Transceiver Tx bias current
sfp-tx-power (dBm)	Transceiver transmitted optical power
sfp-rx-power (dBm)	Transceiver received optical power
eeprom-checksum (good | bad)	Whether EEPROM checksum is correct
eeprom (hex dump)	Raw EEPROM of the transceiver

Example output of an Ethernet status:

Example output of an SFP status:

Detect Cable Problems

A cable test can detect problems or measure the approximate cable length if the cable is unplugged on the other end and there is, therefore, «no-link». RouterOS will show:

• which cable pair is damaged
• the distance to the problem
• how exactly the cable is broken — short-circuited or open-circuited

This also works if the other end is simply unplugged — in that case, the total cable length will be shown.

Here is an example output:

In the above example, the cable is not shorted but “open” at 4 meters distance, all cable pairs are equally faulty at the same distance from the switch chip.

Currently cable-test is implemented on the following devices:

• CCR series devices
• CRS1xx series devices
• CRS2xx series devices
• OmniTIK series devices
• RB450G series devices
• RB951 series devices
• RB2011 series devices
• RB4011 series devices
• RB750Gr2
• RB750UPr2
• RB751U-2HnD
• RB850Gx2
• RB931-2nD
• RB941-2nD
• RB952Ui-5ac2nD
• RB962UiGS-5HacT2HnT
• RB1100AHx2
• RB1100x4
• RBD52G-5HacD2HnD
• RBcAPGi-5acD2nD
• RBmAP2n
• RBmAP2nD
• RBwsAP-5Hac2nD
• RB3011UiAS-RM
• RBMetal 2SHPn
• RBDynaDishG-5HacD
• RBLDFG-5acD
• RBLHGG-5acD

Note: Currently cable-test is not supported on Combo ports.

Stats

Using /interface ethernet print stats command, it is possible to see a wide range of Ethernet-related statistics. The list of statistics can differ between RouterBoard devices due to different Ethernet drivers. The list below contains all available counters across all RouterBoard devices. Most of the Ethernet statistics can be remotely monitored using SNMP and MIKROTIK-MIB.

Property	Description
driver-rx-byte (integer)	Total count of received bytes on device CPU
driver-rx-packet (integer)	Total count of received packets on device CPU
driver-tx-byte (integer)	Total count of transmitted bytes by device CPU
driver-tx-packet (integer)	Total count of transmitted packets by device CPU
rx-64 (integer)	Total count of received 64 byte frames
rx-65-127 (integer)	Total count of received 65 to 127 byte frames
rx-128-255 (integer)	Total count of received 128 to 255 byte frames
rx-256-511 (integer)	Total count of received 256 to 511 byte frames
rx-512-1023 (integer)	Total count of received 512 to 1023 byte frames
rx-1024-1518 (integer)	Total count of received 1024 to 1518 byte frames
rx-1519-max (integer)	Total count of received frames larger than 1519 bytes
rx-align-error (integer)	Total count of received align error events — packets where bits are not aligned along octet boundaries
rx-broadcast (integer)	Total count of received broadcast frames
rx-bytes (integer)	Total count of received bytes
rx-carrier-error (integer)	Total count of received frames with carrier sense error
rx-code-error (integer)	Total count of received frames with code error
rx-control (integer)	Total count of received control or pause frames
rx-error-events (integer)	Total count of received frames with the active error event
rx-fcs-error (integer)	Total count of received frames with incorrect checksum
rx-fragment (integer)	Total count of received fragmented frames (not related to IP fragmentation)
rx-ip-header-checksum-error (integer)	Total count of received frames with IP header checksum error
rx-jabber (integer)	Total count of received jabbed packets — a packet that is transmitted longer than the maximum packet length
rx-length-error (integer)	Total count of received frames with frame length error
rx-multicast (integer)	Total count of received multicast frames
rx-overflow (integer)	Total count of received overflowed frames can be caused when device resources are insufficient to receive a certain frame
rx-pause (integer)	Total count of received pause frames
rx-runt (integer)	Total count of received frames shorter than the minimum 64 bytes, is usually caused by collisions
rx-tcp-checksum-error (integer)	Total count of received frames with TCP header checksum error
rx-too-long (integer)	Total count of received frames that were larger than the maximum supported frame size by the network device, see the max-l2mtu property
rx-too-short (integer)	Total count of the received frame shorter than the minimum 64 bytes
rx-udp-checksum-error (integer)	Total count of received frames with UDP header checksum error
rx-unicast (integer)	Total count of received unicast frames
rx-unknown-op (integer)	Total count of received frames with unknown Ethernet protocol
tx-64 (integer)	Total count of transmitted 64 byte frames
tx-65-127 (integer)	Total count of transmitted 65 to 127 byte frames
tx-128-255 (integer)	Total count of transmitted 128 to 255 byte frames
tx-256-511 (integer)	Total count of transmitted 256 to 511 byte frames
tx-512-1023 (integer)	Total count of transmitted 512 to 1023 byte frames
tx-1024-1518 (integer)	Total count of transmitted 1024 to 1518 byte frames
tx-1519-max (integer)	Total count of transmitted frames larger than 1519 bytes
tx-align-error (integer)	Total count of transmitted align error events — packets where bits are not aligned along octet boundaries
tx-broadcast (integer)	Total count of transmitted broadcast frames
tx-bytes (integer)	Total count of transmitted bytes
tx-collision (integer)	Total count of transmitted frames that made collisions
tx-control (integer)	Total count of transmitted control or pause frames
tx-deferred (integer)	Total count of transmitted frames that were delayed on its first transmit attempt due to already busy medium
tx-drop (integer)	Total count of transmitted frames that were dropped due to the already full output queue
tx-excessive-collision (integer)	Total count of transmitted frames that already made multiple collisions and never got successfully transmitted
tx-excessive-deferred (integer)	Total count of transmitted frames that were deferred for an excessive period of time due to an already busy medium
tx-fcs-error (integer)	Total count of transmitted frames with incorrect checksum
tx-fragment (integer)	Total count of transmitted fragmented frames (not related to IP fragmentation)
tx-carrier-sense-error (integer)	Total count of transmitted frames with carrier sense error
tx-late-collision (integer)	Total count of transmitted frames that made collision after being already halfway transmitted
tx-multicast (integer)	Total count of transmitted multicast frames
tx-multiple-collision (integer)	Total count of transmitted frames that made more than one collision and subsequently transmitted successfully
tx-overflow (integer)	Total count of transmitted overflowed frames
tx-pause (integer)	Total count of transmitted pause frames
tx-all-queue-drop-byte (integer)	Total count of transmitted bytes dropped by all output queues
tx-all-queue-drop-packet (integer)	Total count of transmitted packets dropped by all output queues
tx-queueX-byte (integer)	Total count of transmitted bytes on a certain queue, the X should be replaced with a queue number
tx-queueX-packet (integer)	Total count of transmitted frames on a certain queue, the X should be replaced with a queue number
tx-runt (integer)	Total count of transmitted frames shorter than the minimum 64 bytes, is usually caused by collisions
tx-too-short (integer)	Total count of transmitted frames shorter than the minimum 64 bytes
tx-rx-64 (integer)	Total count of transmitted and received 64 byte frames
tx-rx-64-127 (integer)	Total count of transmitted and received 64 to 127 byte frames
tx-rx-128-255 (integer)	Total count of transmitted and received 128 to 255 byte frames
tx-rx-256-511 (integer)	Total count of transmitted and received 256 to 511 byte frames
tx-rx-512-1023 (integer)	Total count of transmitted and received 512 to 1023 byte frames
tx-rx-1024-max (integer)	Total count of transmitted and received frames larger than 1024 bytes
tx-single-collision (integer)	Total count of transmitted frames that made only a single collision and subsequently transmitted successfully
tx-too-long (integer)	Total count of transmitted packets that were larger than the maximum packet size
tx-underrun (integer)	Total count of transmitted underrun packets
tx-unicast (integer)	Total count of transmitted unicast frames

For example, the output of Ethernet stats on the hAP ac2 device:

Источник

Столкнулся с проблемой на MikroTik – ошибки в логах и отсутствие интернета на порту Ether1. Стал искать ответ в интернете, но решения не нашел. Существует предположение, что проблема появления ошибки interface,warning ether1 fcs error on link связана с наводками от кабелей 220v.

К сожалению, проблема не имеет стойкого графика появления. Цикл между ошибками разный, может быть неделя, а может быть и пару месяцев. Помогает либо полная перезагрузка MikroTik или отключение порта. До текущего момента мы делали это вручную, после был найден скрипт. К сожалению автора скрипта, найти не удалось. Но раз ошибка популярная, то думаю не лишним будет его указать, как решение проблемы.

На микротике RB951G-2HnD установлена последняя RouterOS 6.19.
Провайдер (Ростелеком) подключен в первый порт роутера, скорость соединения в микротике определяется, как 100 full duplex.

При поднятии PPPoE соединения с провайдером теряется около 40% пакетов через это соединение. Пробовал ping 8.8.8.8 и ping ya.ru и просто открывал сайты через http.

Важно, что при использовании старого оборудования (сервер на Intel Atom) таких потерь не обнаруживается. Пробовал подключать ноутбук с Ubuntu напрямую тоже все работает без проблем.

Что делал:
Менял MTU PPPoE соединения и flow control Ethernet port1.
Менял роутер для исключения не работоспособности железа.
Менял физический порт роутера на 5й эффект сохранялся.
На форуме микротика нашел схожую проблему, которая якобы исправлена в версии 6.11 RouterOS.

Также как и предложено на форуме микротика включил между микротиком и провайдером Dlink-DGS 1005d и потери пакетов исчезли.

Как исправить проблему с потерей пакетов? Хочу подключать напрямую, а не через доп. оборудование.

суббота, 27 декабря 2014 г.

Значения счетчиков ошибок

Не всегда понятно что может означать та или иная ошибка при передаче. Ниже моя попытка объяснить значение счетчиков ошибок, которые регистрирует коммутатор.

Счетчики ошибок при получении кадров (RX):

CRC Error

Counts otherwise valid packets that did not end on a byte (octet) boundary.

Счетчик ошибок контрольной суммы (CRC). В свою очередь, является суммой счетчиков Alignment Errors и FCS Errors.
FCS (Frame Check Sequence) Errors — ошибки в контрольной последовательности кадра. Счетчик регистрирует кадры с ошибками FCS, при этом кадры имеют корректный размер (от 64 до 1518 байт) и получены без ошибок кадрирования или коллизий.
Alignment Errors — ошибки выравнивания (некорректной длины кадра). Счетчик регистрирует кадры с ошибками FCS, при этом кадры имеют корректный размер (от 64 до 1518 байт), но были получены с ошибками кадрирования.
В случае, если кадр был классифицирован как имеющий ошибку Alignment Error, счетчик FCS при этом не увеличивается. Иными словами, инкрементируется либо счетчик FCS либо Aligment, но не оба сразу.

UnderSize

The number of packets detected that are less than the minimum permitted packets size of 64
bytes and have a good CRC. Undersize packets usually indicate collision fragments, a normal
network occurrence.

Счетчик кадров с правильной контрольной суммой и размером менее 64 байт. Такие кадры могут возникать в результате коллизий в сети.

OverSize

Counts valid packets received that were longer than 1518 octets and less than the
MAX_PKT_LEN. Internally, MAX_PKT_LEN is equal to 1536.

Счетчик кадров с правильной контрольной суммой, размер которых превышает 1518 байт, но не превышает 1536 байт — внутреннего максимального значения кадра.

Fragment

The number of packets less than 64 bytes with either bad framing or an invalid CRC. These
are normally the result of collisions.

Счетчик кадров с неправильной контрольной суммой или структурой кадра и размером менее 64 байт. Такие кадры могут возникать в результате коллизий в сети.

Jabber

Counts invalid packets received that were longer than 1518 octets and less than the
MAX_PKT_LEN. Internally, MAX_PKT_LEN is equal to 1536.

Счетчик кадров с неправильной контрольной суммой, размер которых превышает 1518 байт, но не превышает 1536 байт — внутренного максимального значения кадра.

Счетчик ошибок при отправке кадров (TX):

Excessive Deferrral

Counts the number of packets for which the first transmission attempt on a particular
interface was delayed because the medium was busy.

Счетчик кадров, первая попытка отправки которых было отложена из-за занятости среды передачи.

CRC Error

Counts otherwise valid packets that did not end on a byte (octet) boundary.

Счетчик ошибок контрольной суммы (CRC). На практике никогда не увеличивается.

Late Collision

Counts the number of times that a collision is detected later than 512 bit-times into the
transmission of a packet.

Счетчик случаев когда коллизия обнаруживалась после передачи первых 64 байт (512 бит) кадра.

Excessive Collision

Excessive Collisions. The number of packets for which transmission failed due to excessive
collisions.

Счетчик кадров, отправка которых не удалась из-за чрезмерного количества колизий.

Single Collision

Single Collision Frames. The number of successfully transmitted packets for which
transmission is inhibited by more than one collision.

Счетчик успешно отправленных кадров, передача которых вызвала более одной коллизии.

Collision

Моно добавить, что на практике RX CRC обычно является результатом деградации среды передачи (медный кабель или оптоволокно), а TX-коллизии — результатом неправильного согласования скорости соединения, например half-линка.

Неплохая расшифровка значений счетчиков приведена тут.