I have checked and after more than 24 hours, I still have my CPU load at 0 with https enabled.
I have also searched the forums and see that some people have, in the past, had issues that seem similar, and this goes back many builds. However, many are also not seeing this.
I enabled https to check if that alone was causing the issue. It appears that it isn't. Thanks Geetek for checking this on your system too.
One thing I did note in your settings was that DynDNS. It seems to me that some people have experienced problems with this. You might want to do a search on that. However the problem also could be related to encryption.
I will enable WPA2-AES on this router. Would it be possible to disable DynDns on yours, and then we can see if this makes a difference for either of us?
EDIT: WPA2-AES seems to be settled. I am using this router now (so there is some WAN traffic) and the gui is still showing O.O.O for load. It did rise to .2 at some point.
cybermcm could you also please confirm your load numbers? I believe Redhawk reported a problem with top output in the past and busybox showing 100% load incorrectly. _________________ SIG:
I'm trying to teach you to fish, not give you a fish. If you just want a fish, wait for a fisherman who hands them out. I'm more of a fishing instructor.
LOM: "If you show that you have not bothered to read the forum announcements or to follow the advices in them then the level of help available for you will drop substantially, also known as Murrkf's law.."
Posted: Sat Nov 07, 2009 16:28 Post subject: Re: old httpd 100% bug back again?
GeeTek wrote:
Since you mentioned that it takes several days for the symptom to exhibit, do you think it may be related to inbound WAN traffic ?
For now, I've no idea if its related to the inbound WAN traffic but possible. Today I'm having a lot of P2P traffic and the problem comes a few times a day!
Murrkf wrote:
cybermcm could you also please confirm your load numbers? I believe Redhawk reported a problem with top output in the past and busybox showing 100% load incorrectly.
What do you exactly mean with "load numbers"?
I'm using build 12548M at the moment where top and web gui show 100% CPU load.
I'll try and disable DynDNS and will report back if this changes anything...
Posted: Sat Nov 07, 2009 16:57 Post subject: Re: old httpd 100% bug back again?
cybermcm wrote:
Today I'm having a lot of P2P traffic and the problem comes a few times a day!
Hold the phone....Turn THAT off. P2P does cause extreme load on the CPU. (and would be what Geetek referred to as Wan Traffic ). I would turn off just that to start. Leave the DynDNS on.
In top corner of the webgui, like what I posted, you will see three numbers. Those show the CPU load average over time. (That is why I posted them, BTW). I would like to know the numbers that you see there.
All future posts of zorganizer will be deleted. His posts are offensive, immature and basically the ramblings of an a-hole with a fragile ego. He has a history of this. He offers nothing but personal attacks and complaints in an effort to get others to do what is beyond his personal intellectual ability. We don't tolerate that around here. With him gone, we can focus on finding a solution to the issue you are having. _________________ SIG:
I'm trying to teach you to fish, not give you a fish. If you just want a fish, wait for a fisherman who hands them out. I'm more of a fishing instructor.
LOM: "If you show that you have not bothered to read the forum announcements or to follow the advices in them then the level of help available for you will drop substantially, also known as Murrkf's law.."
Posted: Sat Nov 07, 2009 17:07 Post subject: Re: old httpd 100% bug back again?
Murrkf wrote:
Hold the phone....Turn THAT off. P2P does cause extreme load on the CPU. (and would be what Geetek referred to as Wan Traffic ). I would turn off just that to start.
In top corner of the webgui, like what I posted, you will see three numbers. Those show the CPU load average over time. I would like to know the numbers that you see there.
The bug also appeared when there was no P2P traffic. Also reducing the maximum ports nor the timeouts did help. I also tried to close my P2P app after noticing the bug but the CPU load didn't change. So for me it doesn't seem to be related to the problem.
I'll report the 3 numbers when the bug appears next time, thx for explaining...
Posted: Sat Nov 07, 2009 17:44 Post subject: Re: old httpd 100% bug back again?
cybermcm wrote:
The bug also appeared when there was no P2P traffic. Also reducing the maximum ports nor the timeouts did help. I also tried to close my P2P app after noticing the bug but the CPU load didn't change.
What are the three numbers now? And, until we have this sorted out, could you keep the p2p off. The load might not go down right away. What do you have your timeouts set to?
And, how long does it take for them to reach high load? I have wpa2-aes and the load numbers are still low. .09,.08,.01
If you don't think it is p2p, go to the DynDNS next. Shut it down. Also, you should have noted that there is a link to the dynDNS issues in the peacock announcement. Did you read those linked posts? Seems to me they might be relevant. _________________ SIG:
I'm trying to teach you to fish, not give you a fish. If you just want a fish, wait for a fisherman who hands them out. I'm more of a fishing instructor.
LOM: "If you show that you have not bothered to read the forum announcements or to follow the advices in them then the level of help available for you will drop substantially, also known as Murrkf's law.."
Posted: Sat Nov 07, 2009 19:24 Post subject: Re: old httpd 100% bug back again?
Quote:
What are the three numbers now? And, until we have this sorted out, could you keep the p2p off. The load might not go down right away. What do you have your timeouts set to?
The numbers change constantly, now they are 0.66, 0.36, 0.13, but they are increasing as soon as I use the web interface (no 100% bug right now)
For the the moment I'll continue to use P2P to force the bug again.
Ports and timeout are at standard, 4096 ports, timeout 3600 and 120.
If the bug appears and I shut down the P2P app even 1 hour later the CPU load is still at 100%.
Quote:
And, how long does it take for them to reach high load? I have wpa2-aes and the load numbers are still low. .09,.08,.01
I can't reproduce the bug with a special action. Sometimes it works for a day and sometimes the bug appears a couple of hours after a reset...
Quote:
If you don't think it is p2p, go to the DynDNS next. Shut it down. Also, you should have noted that there is a link to the dynDNS issues in the peacock announcement. Did you read those linked posts? Seems to me they might be relevant.
I disabled DynDNS and will watch the status. I didn't read about DynDNS issues, I'll do that right away...
Posted: Sat Nov 07, 2009 19:41 Post subject: Re: old httpd 100% bug back again?
cybermcm wrote:
For the the moment I'll continue to use P2P to force the bug again.
We actually want to see what makes it stop, not try to create it. P2P does increase load. I suggest you stop it until you get this resolved.
However, I am thinking DynDNS might be the issue. My load is not getting past about .20. Right now, when using the router, the highest reading is .08. It's been going for about 2 days like this. _________________ SIG:
I'm trying to teach you to fish, not give you a fish. If you just want a fish, wait for a fisherman who hands them out. I'm more of a fishing instructor.
LOM: "If you show that you have not bothered to read the forum announcements or to follow the advices in them then the level of help available for you will drop substantially, also known as Murrkf's law.."
Joined: 06 Jun 2006 Posts: 3763 Location: I'm the one on the plate.
Posted: Sat Nov 07, 2009 20:12 Post subject: Re: old httpd 100% bug back again?
Murrkf wrote:
P2P does increase load. I suggest you stop it until you get this resolved.
Another thing to consider on P2P is that even though you shut down the P2P client on your computer, the P2P pools that you have been a member of will continue to pound your IP address for days trying to establish connections.
cybermcm, when you catch the load at 100%, unplug the WAN port and see if the CPU load drops back. _________________ http://69.175.13.131:8015 Streaming Week-End Disco. Station Ripper V 1.1 will do.
I'm also experience the 100% CPU load.
My router is Asus WL-500G Premium v2.
Firmware info:
v24 preSP2 (Build13064)
DD-WRT v24-sp2 (10/10/09) mega
Code:
Mem: 13600K used, 15612K free, 0K shrd, 1840K buff, 5784K cached
CPU: 34.5% usr 65.4% sys 0.0% nic 0.0% idle 0.0% io 0.0% irq 0.0% sirq
Load average: 1.11 1.06 0.88 4/24 22674
PID PPID USER STAT VSZ %MEM %CPU COMMAND
10899 1 root R 2980 10.1 99.4 httpd -S
22673 22672 root R 1516 5.1 0.4 top
22670 211 root S 1936 6.6 0.0 dropbear -b /tmp/loginprompt -r /tmp/r
748 1 root R 1924 6.5 0.0 upnp -D -W vlan1
211 1 root S 1872 6.4 0.0 dropbear -b /tmp/loginprompt -r /tmp/r
14 1 root S 1836 6.2 0.0 watchdog
1 0 root S 1800 6.1 0.0 /sbin/init noinitrd
93 1 root S 1796 6.1 0.0 resetbutton
852 1 root S 1792 6.1 0.0 wland
220 1 root S 1792 6.1 0.0 ttraff
597 1 root S 1792 6.1 0.0 process_monitor
253 1 root S 1700 5.8 0.0 nas -P /tmp/nas.wl0lan.pid -H 34954 -l
22672 22670 root S 1528 5.2 0.0 -sh
635 1 root S 1152 3.9 0.0 inadyn --input_file /tmp/ddns/inadyn.c
177 1 root S 1148 3.9 0.0 dnsmasq --conf-file=/tmp/dnsmasq.conf
868 1 root S 1032 3.5 0.0 udhcpc -i vlan1 -p /var/run/udhcpc.pid
632 1 root S 1028 3.5 0.0 cron
3 1 root SWN 0 0.0 0.0 [ksoftirqd_CPU0]
10 1 root SW 0 0.0 0.0 [mtdblockd]
19211 635 root Z 0 0.0 0.0 [sh]
Disabled:
telnet, http
Enable:
https, ssh
The problem appears in one day. Yesterday i had no p2p traffic, today i had p2p traffic.
I'm also unable to get in webgui of the router, so i connect with putty(ssh) to the router and kill httpd -S an than manual start the httpd -S. After that router is working ok for 1 day.
I'll disable ssh and do the 30/30/30 reset. I'll report what will happend.
Posted: Sun Nov 08, 2009 7:38 Post subject: my experience with this problem
I have a WNDR3300 with a BCM4704@264Mhz. I had this very same problem over a broad range of firmwares, though I am now using newd eko 13000.
Sometimes it would be up for a few days, sometimes hours, but ultimately my https process would always crash.
I tried drastically reducing my TCP connection timeout times, to no avail.
Then I changed my https port to a nonstandard one. Problems went away. Been running a month trouble free. Something from the WAN (malicious or otherwise) seems to have been the culprit. At least in my case.
For what it's worth, I use DynDNS, run a VPN server, have ssh enabled, WPA2-AES on two radios, and have torrented over 150GB in that month. I was worried that my router was running out of RAM, as it has only 16MB, but consistently I always have 3-4MB cached even when things get busy. My CPU load is generally low (<.2) even with heavy torrenting. Bandwidth monitoring in real time causes more spikes than anything else. So I do not think the issue is related to router load. At least not in my case. But I ran top and isolated my problem to 100% CPU in the httpd -s process. It sounds like some posts may be talking about a (possibly) different process crashing.
Let me know if any of you have tried changing the port, and if that solves it for you as well.
This time DynDNS was turned off, so this can't be the problem (first I though it my be related to my DynDNS password with special character "$")
There was no P2P traffic.
I tried to disconnect the WAN cable, waited 5 min, the CPU load won't get lower.
Next thing I'll try is changing the https port (bbalaban hint) but this option is rather useless for me because from my work network I'm only able to connect to port 80 and 8080...
Joined: 06 Jun 2006 Posts: 3763 Location: I'm the one on the plate.
Posted: Sun Nov 08, 2009 12:12 Post subject:
cybermcm wrote:
Next thing I'll try is changing the https port (bbalaban hint) but this option is rather useless for me because from my work network I'm only able to connect to port 80 and 8080...
If the alternate port fixes it then you can port translate in the router until a better solution is found.
Edit - Nice ! When was the "from" port added to the GUI port forward rules ? _________________ http://69.175.13.131:8015 Streaming Week-End Disco. Station Ripper V 1.1 will do.