Posted: Wed Feb 03, 2016 5:58 Post subject: Odd QOS behavior on D-Link DIR-825 rev. b
I searched and could not find information on my particular problem.
I currently have ~ 50 mbps down and ~ 5 mbps up with my current ISP. This was tested through speedtest.net on a laptop hard wired to the modem and no other connections. I am trying to limit my own connection along with another desktop (both Wi-Fi connections) in the home through QOS as we both use most of the bandwidth.
I have done this in the past with satisfying results. I am currently running DD-WRT v3.0-r29002 std. I have tried a few other firmware versions, but keep having the same problem.
The QOS settings (MAC address Manual Priority with max upload and download set as desired) appear to take affect as normal, but upon testing the connection through YouTube and Speedtest.net, it becomes apparent that I missed something. The download speed tests at 20 mbps (as set in QOS) until about 3/4 through the test and the speed drops drasticly to unusable speeds (it appears as a drop on the line graph, but continues to show 15 mbps even though it clearly is not.). I can refresh the page just fine and run the test again with the same result. I don't recall getting an upload speed result since the download test fails. On YouTube, videos begin to load quickly a in short spurt, load a tiny bit more (slowly) and then another bit loads in a quick spurt again.
QoS is pretty buggy, if you use layer7 services, they most likely fail to work: http://svn.dd-wrt.com/ticket/5024
Port based priority seems to work fine though... _________________ Atheros: TL-WR1043ND-v2 @ r33492 std
Ok, now we're getting into some really weird findings. I just conducted an experiment with VERY interesting results. My CPU usage did not exceed 20% and I guess the memory being used is only 38% (I thought it was 94%, but that's the total that's available for use).
I turned on QOS again without changing anything and ran tests at speedtest.net once again. This time, however, I had dd-wrt pulled up on a separate browser window to monitor CPU usage. I continuously hit refresh on that window and you know what? The test ran PERFECTLY! If I run the test without refreshing the other window, the test fails about 3/4 through.
I'm not sure what to make of this, but perhaps it gives someone some insight on what the problem is?
Hmm, did you do a reset after upgrading? AFAIK sometimes it's necessary, depends on what changes has been made in the firmware. There was a discussion about it lately and conclusion is to reset if you experience any new problems.
Status page refreshes automagically (by default every 3s, can be changed) and it uses less resources than manual refresh because it doesn't have to generate/load whole page but only values to update. It would be really weird for the router to work better under higher load... _________________ Atheros: TL-WR1043ND-v2 @ r33492 std
Well, I thought you were on to something there. I had my hopes up, but the problem persists.
After doing the 30/30/30 reset, should i not be able to log into the interface at 192.168.1.1 while the power light is blinking? I can after restarting the router, but I seem to recall logging in just after the final 30 second procedure without rebooting in the past.
Well, that seems to have been a waste of time, with the exception of one discovery. After the process of flashing oem firmware and back to dd-wrt, I tested the connection with QOS enabled. SAME RESULTS. However, I did notice something odd. My blue power light flashes "pink" (combination of yellow and blue led) for a moment shortly after the download speeds drop off.
Seriously, is there still hope or am I looking at shelling out $$$ for a replacement here?
What might some of you suggest if a replacement is needed? I've done a fair bit of research and it seems all routers have a limited life span of ~ 2 years. That being the case, I can't really justify spending more than $100 U.S.
Yes I currently have "Per User Default Limits", but only for testing purposes. The issues I've described in my previous posts are the same even if I set up MAC priority (from the device with the MAC address) and disable user default limits.