AP:
MK112 with prism 2,5 card. 12 connectred clients. And latency si around 2-6 ms. It is very cool but not for long time. A rest of time the latency is around 150 - 400 ms.
If latency is 2-6 ms the throughput could be higher with no roblem. 1200 / 500 kbps
If latency is high the throughput coldn`t be higher than 300-400 kbps.
Defauld forward is disabled, broadcast is filtered with netbios, too.The connections are limited for 150 / user!
And the are a lot of droped packets!I tryed to change the frequency too. It is the same.
I tryed change ROS version. I tryed beta v. too. It is the same!
This problem is not only on thi one AP. I thought:the problem is witch cm9 wlan card.I changed it for Prism 2.5. It is the same.
Please help!!!!!
Re: Abnormal latency on APs
AP:
MK112 with prism 2,5 card. 12 connectred clients. And latency si around 2-6 ms. It is very cool but not for long time. A rest of time the latency is around 150 - 400 ms.
If latency is 2-6 ms the throughput could be higher with no roblem. 1200 / 500 kbps
If latency is high the throughput coldn`t be higher than 300-400 kbps.
Defauld forward is disabled, broadcast is filtered with netbios, too.The connections are limited for 150 / user!
And the are a lot of droped packets!I tryed to change the frequency too. It is the same.
I tryed change ROS version. I tryed beta v. too. It is the same!
This problem is not only on thi one AP. I thought:the problem is witch cm9 wlan card.I changed it for Prism 2.5. It is the same.
Please help!!!!!
LAN and Wlan interfaces are bridged!
Where are you seeing packet drops ?Where are you seeing packet drops? How do you measure it?
What's you configuration?
Some pings are droped if i`am trying ping connected client on AP! around 20 %
With ping command on the command line
What's you configuration?
What kind of configuration you mean? I use 2.4 GHz B
cpu load is around 20 %What's the load at the time? What's the response time? Exactly what kind of client are you pinging? What is the AP wireless configuration?
ping for ethernet interface is all right!
the clients are ovislink 1120,5450,5460
Code: Select all
name=" XXXXXXXX " mtu=1500 mac-address=00:02:6F:xx:xx:xx arp=enabled
disable-running-check=no interface-type=Prism prism-cardtype=100mW
radio-name="XXXXXXXX" mode=ap-bridge ssid="XXXXXXXX" area=""
frequency-mode=manual-txpower country=rusia antenna-gain=0
frequency=2427 band=2.4ghz-b scan-list=default rate-set=default
supported-rates-b=1Mbps,2Mbps basic-rates-b=1Mbps,2Mbps
max-station-count=2007 tx-power-mode=default
periodic-calibration=enabled periodic-calibration-interval=60
dfs-mode=none antenna-mode=ant-a wds-mode=disabled
wds-default-bridge=none wds-default-cost=100 wds-cost-range=50-150
wds-ignore-ssid=no update-stats-interval=disabled
default-authentication=yes default-forwarding=no default-ap-tx-limit=0
default-client-tx-limit=0 hide-ssid=no security-profile=default
disconnect-timeout=3s on-fail-retry-time=300ms compression=no
allow-sharedkey=no
Re: Abnormal latency on APs
You can get some helpful information by doubleclcking on an entry in the wireless/registration table and looking at the Statistics tab regarding what is happening below layer 3.AP:
And the are a lot of droped packets!I tryed to change the frequency too. It is the same.
Pay close attention to Tx/Rx Frames vx. TX/Rx HW Frames. For a good connection, they will be identical, even if in the millions or higher.
Further, local interference in a neighborhood may mean different service levels across the coverage area of the AP (if, e.g., said AP covers multiple neighborhoods), so optimizing cell size (and thus your exposure to multiple sources of interference) can be an issue.
Solve the lower layer problems and the layer 3 starts working magically unless you are dropping the packets yourself.
regards/ldv
Re: Abnormal latency on APs
If you cycle between times of good latency and poor latency, look for the cause of the cycle.MK112 with prism 2,5 card. 12 connectred clients. And latency si around 2-6 ms. It is very cool but not for long time. A rest of time the latency is around 150 - 400 ms.
A single cycle (a single occurrence) can be more difficult to diagnose than a repetitive cycle, but you wind up looking at the data to find out what changed (changes) at the inflection point(s) or executing a series of SWAGs.
Turn on wireless logging and look at the log file as well as the Statistics tab for particular connections under wireless/registration-table.
best regards/ldv
Re: Abnormal latency on APs
The cycle between times of good latency and poor latency is not equal.Usualy the problems with poor latency are at night.If you cycle between times of good latency and poor latency, look for the cause of the cycle.MK112 with prism 2,5 card. 12 connectred clients. And latency si around 2-6 ms. It is very cool but not for long time. A rest of time the latency is around 150 - 400 ms.
A single cycle (a single occurrence) can be more difficult to diagnose than a repetitive cycle, but you wind up looking at the data to find out what changed (changes) at the inflection point(s) or executing a series of SWAGs.
Turn on wireless logging and look at the log file as well as the Statistics tab for particular connections under wireless/registration-table.
best regards/ldv
the severance betwen Tx/Rx Frames vx. TX/Rx HW Frames are monimal. RX frames are equal. TX is little different ...Very little
Re: Abnormal latency on APs
Were the ping times good or poor when you took that lab note?the severance betwen Tx/Rx Frames vx. TX/Rx HW Frames are monimal. RX frames are equal. TX is little different ...Very little
Whether good or bad, did you make a second lab note regarding the opposite condition?
If so, what was that result?
regards/ldv
Re: Abnormal latency on APs
Were the ping times good or poor when you took that lab note?the severance betwen Tx/Rx Frames vx. TX/Rx HW Frames are monimal. RX frames are equal. TX is little different ...Very little
Whether good or bad, did you make a second lab note regarding the opposite condition?
If so, what was that result?
regards/ldv
Sorry, I don`t understand what you said
Re: Abnormal latency on APs
the severance betwen Tx/Rx Frames vx. TX/Rx HW Frames are monimal. RX frames are equal. TX is little different ...Very little
Were the ping times good or poor when you took that lab note?
Whether good or bad, did you make a second lab note regarding the opposite condition?
If so, what was that result?
1. Please reply with lab notes regarding:Sorry, I don`t understand what you said
Tx/Rx Frames vs. Tx/Rx HW Frames
1a). under good conditions (good ping times)
and
1b). under poor conditions (poor ping times).
2. Please turn on/enable wireless under Winbox/system/logging and check said syslog for interesting (i.e., not nominal) entries during times of poor conditions (poor ping times).
regards/ldv
That is an old myth since before Atheros cards were plentiful.I heard so the best effort for AP whit MK is mini PCI card with prism 2,5 chipset.
And ORiNOCO 11b cards reset whenever they are hit with short-rf preamble packets. And if you buy a batch of them, they fall apart at roughly the same time (indicating they are designed to break down).
Hello,
I have smiliar problems on my network.
I am using RB-532A everywhere as APs and different platforms as a clients, including MT loaded on wrapboards. I have point-to-point 5GHz connection between main nodes of the network and they are performing flawlessly.
I am using static routing everywhere, no bridging. For the clients I'm using 2.4GHz frequency range.
Most of my clients are constantly streaming video at ~250kbit/s as I am providing monitoring service. I would only have three or four clients on an AP. The problem occurs when I have more than two clients on one AP.
I have tried everything I could think of. The signal strength is always above -70dbm. There is no one else using the same or neighbouring frequencies.
Torch is only showing the traffic I know about (I tried torching while poor as well as good network performance).
I have found, however one thing that is not right according to one of the previous posts. THe Tx/Rx Frames and TX/Rx HW Frames are totally different on some of the interfaces.
What does it mean and how can I fix it?
Thanks in advance for your help,
Regards
I have smiliar problems on my network.
I am using RB-532A everywhere as APs and different platforms as a clients, including MT loaded on wrapboards. I have point-to-point 5GHz connection between main nodes of the network and they are performing flawlessly.
I am using static routing everywhere, no bridging. For the clients I'm using 2.4GHz frequency range.
Most of my clients are constantly streaming video at ~250kbit/s as I am providing monitoring service. I would only have three or four clients on an AP. The problem occurs when I have more than two clients on one AP.
I have tried everything I could think of. The signal strength is always above -70dbm. There is no one else using the same or neighbouring frequencies.
Torch is only showing the traffic I know about (I tried torching while poor as well as good network performance).
I have found, however one thing that is not right according to one of the previous posts. THe Tx/Rx Frames and TX/Rx HW Frames are totally different on some of the interfaces.
What does it mean and how can I fix it?
Thanks in advance for your help,
Regards
I'll bet you $100 that John meant for you to key inSystem reboot helps no more than 5 minutes... Reboot is not solving !As usual, your rate settings are all messed up. A /system reset is probably easiest.
/system reset
at the CLI prompt.
John's from Memphis IIRC and may therefore like his BBQ sauce to have a little vinegar and I'm almost as certain that John would have written
/system reboot
if his intent was to get you to reboot the box.
You get the CLI command prompt if you SSH to the box or click on "New Terminal" in Winbox.
ACHTUNG ALLESLOOKENSPEEPERS!
Invoking '/system reset' at the CLI command prompt will return your box to a factory default configuration. Please be prepared to take control again and reconfigure your box as you did when you first received the box.
regards/ldv
I'd like to suggest that you:The sameWhat is the configuration now then?
. export your configuration
. redact any proprietary information
. post it on this forum
. expect multiple experienced members of the MikroTik community, MikroTik itself and its dealers, to react with informed suggestions.
regards/ldv
Interesting thread, I have recently run into a couple of situations I can fingure out. 1st we recently upgraded a water tower by removing a Tranzeo AP 4500 with a Mt 532 and SR2 card. We are using 64bit WEP, Mac auth, hidden SSID and about 10 to 30 seconds after enabling the wireless interface the latency goes over 2000ms and the AP dies. After a couple of days trouble shooting I did a few packet sniffs. What I found is mostly probe requests. Most of the requests were from my clients. It appeared they were not able to register. The MAC's having problems were 00:60:B3 and 00:0B:6B. These are Tranzeo clients radios. In desperation I switched back to the Tranzeo AP and all if fine. It appears to me we still have an issue with compatability with some radio chipsets. Any thoughts?
2nd Can someone define the differance between Tx/RX Rate, Frames, and HW Frames. In my experience the Tx/Rx Rate and Frames are usually sabout the same, The HW Frames are usually higher. Does the differace represent the dropped and retransmitted packets?
I know I said 2 but here is a 3rd. Some of the fields in int wir mon aren't documented. For example from the station side we have p-throughput. I assume it is an indication of packet per but it is running aroung 29000 with traffic running around 5Mbps.
Hope this isn't to much.
2nd Can someone define the differance between Tx/RX Rate, Frames, and HW Frames. In my experience the Tx/Rx Rate and Frames are usually sabout the same, The HW Frames are usually higher. Does the differace represent the dropped and retransmitted packets?
I know I said 2 but here is a 3rd. Some of the fields in int wir mon aren't documented. For example from the station side we have p-throughput. I assume it is an indication of packet per but it is running aroung 29000 with traffic running around 5Mbps.
Hope this isn't to much.
I'm not very confident this will be helpful to you, but some folks running Tranzeos wind up trying to solve certain problems by adjusting RTS and fragmentation thresholds to lower values. Lonnie N. has written quite a bit on his forum regarding the evils of low numbers which are thought to solve, e.g., the hidden node problem, but cause others. This situation can normally be detected by using 1500 byte pings.These are Tranzeo clients radios. In desperation I switched back to the Tranzeo AP and all if fine. It appears to me we still have an issue with compatability with some radio chipsets. Any thoughts?
Some strange symptoms with low values are:
1. IE will display ftp://ftp.utexas.edu but not http://www.google.com, e.g.
2. same behavior with Opera
3. 1500 byte pings don't get echoed.
regards/ldv
Post the version of the RouterOs used and the firmware of the RB112.
I have the same issue in the past, I have reduced the problem as follow:
1) update the firmware from 1.3 to the last 2.6
2) update the ROUTEROS to last
3) I have reduced the upload from the clients to internet. (for 50 customers from 2 Mbit to 600 kbits)
with this modification the download increase to 8 Mbit for sector.
In some sector I have switched from b-g to only b since I have GSM operator in the same roof.
I have the same issue in the past, I have reduced the problem as follow:
1) update the firmware from 1.3 to the last 2.6
2) update the ROUTEROS to last
3) I have reduced the upload from the clients to internet. (for 50 customers from 2 Mbit to 600 kbits)
with this modification the download increase to 8 Mbit for sector.
In some sector I have switched from b-g to only b since I have GSM operator in the same roof.
How can I reduce upload on ap. What is the best effort?Post the version of the RouterOs used and the firmware of the RB112.
I have the same issue in the past, I have reduced the problem as follow:
1) update the firmware from 1.3 to the last 2.6
2) update the ROUTEROS to last
3) I have reduced the upload from the clients to internet. (for 50 customers from 2 Mbit to 600 kbits)
with this modification the download increase to 8 Mbit for sector.
In some sector I have switched from b-g to only b since I have GSM operator in the same roof.
Hi maxfava,3) I have reduced the upload from the clients to internet. (for 50 customers from 2 Mbit to 600 kbits)
with this modification the download increase to 8 Mbit for sector.
is that 600 Kbps cumulative across all 50 customers or per customer (counting on a statistical distribution)?
Care to share your queue code?
great regards/ldv