Livestat module use locks up computer

[willyw]

Windrop 1.6.19
OS: CYGWIN_NT-5.2 1.5.25(0.156/4/2)
Tcl library: ./lib/tcl8.5
Tcl version: 8.5.2 (header version 8.5.2)

We have two bots. Both use same version. Both on same computer. Different installations though - different directories, conf files, etc.

First bot has been online for quite a while. No problems with Livestats module. Seems to display fine, when accessed via web.

Second bot has been online for less time. Have not noticed any problems until now though.
Decided to implement Livestats on the second bot. Displays some stats ok, but locks up on Top30. Not just the bot... but it locks up the whole computer.

The port in use for Livestats is unique - not used for anything else. And we tried changing it a time or two... same result.

In partyline, we noticed something.
Example:

<botnickt> [14:54] xx.xx.xxx.xxx: GET /channame/onchan/ HTTP/1.1
<botnick> [14:54] net: connect! sock 11
<botnick> [14:54] xxx.xxx.x.x: GET /favicon.ico HTTP/1.1
<botnick> [14:54] net: connect! sock 11
<botnick> [14:54] xxx.xxx.x.x: GET /favicon.ico/ HTTP/1.1
<botnick> [14:54] net: connect! sock 11
<botnick> [14:54] xx.xx.xx.xxx: GET /channame/users/ HTTP/1.1
<botnick> [14:54] net: connect! sock 11
<botnick> [14:54] xxx.xxx.x.x: GET /favicon.ico HTTP/1.1
<botnick> [14:54] net: connect! sock 11
<botnick> [14:54] xxx.xxx.x.x: GET /favicon.ico/ HTTP/1.1
<botnick> [14:54] net: connect! sock 11
<botnick> [14:54] net: connect! sock 11
<botnick> [14:54] xx.xx.xx.xxx: GET /channame/users/someuser/ HTTP/1.1
<botnick> [14:55] net: connect! sock 11
<botnick> [14:55] xx.xx.xxx.xxx: GET /channame/misc/ HTTP/1.1
<botnick> [14:55] net: connect! sock 11


Perhaps that line, "net: connect! sock 11" is a clue, as it does not appear in partyline on the first bot.


Also, when doing .restart , have noticed the same line:
net: connect! sock 11
as the bot loads. ( The number varies.)
We do not see this when .restart the first bot.


We would very much like to be able to use Livestats on this ( the second ) bot. It is most frustrating, since it works fine on the first bot. I've been over the config, and can't discern any difference that might be causing it.

Any ideas as to what we might do to get it going?

If you need more info and/or test and test results, please describe what you need, and I'll get it.


Thanks

[willyw]

No replies?

Perhaps I posted in the wrong area? If so, sorry - and please let me know.

[nml375]

The post is good (although being a windrop *j*)

I don't run eggies under windows myself, but with the NT-kernel, there generally shouldn't be any way for a userspace process (your eggdrop) to crash the whole system. This would usually indicate a bad driver is to blame.
Since eggies don't use fancy graphics or advanced 3d sounds, what's left is pretty much your network card drivers.

Unfortunately, I can't be of much more help than that.

[willyw]

First, thank you for replying.

The post is good (although being a windrop *j*)

I don't run eggies under windows myself, but with the NT-kernel, there generally shouldn't be any way for a userspace process (your eggdrop) to crash the whole system. This would usually indicate a bad driver is to blame.
Since eggies don't use fancy graphics or advanced 3d sounds, what's left is pretty much your network card drivers.

Unfortunately, I can't be of much more help than that.

A couple questions though, if I may:

If a driver is the cause, wouldn't the problem then appear on both bots?

Other question: do you (or anyone else reading along) know what that entry " net: connect " means? What is that telling me?
It only appears on the bot that has the problem, so I'm zeroing in on it for now.

Thanks

[nml375]

The error might not appear on both bots even if it is a driver issue, as there might be slight changes within the configs that makes your bot utilize the network layer differently, or perhaps you always start one bot before the other (suggesting a resource issue with the driver).

"net: connect!" is not an error message, but rather a debug notice letting you know there was an incoming connection on socket 11 (most likely the socket created to handle the livestats requests). This again suggests you've got different settings on your bot, which might be enough to "trip the wire" and crash the system..

[willyw]

The error might not appear on both bots even if it is a driver issue, as there might be slight changes within the configs that makes your bot utilize the network layer differently, or perhaps you always start one bot before the other (suggesting a resource issue with the driver).

Both are started by shortcuts to batch files, placed in the startup folder.
Chances are, one is always started first.

Would simply NOT starting the working bot be a good test?
In other words, start only one bot, and that bot being the one that is exhibiting the problem?

"net: connect!" is not an error message, but rather a debug notice letting you know there was an incoming connection on socket 11 (most likely the socket created to handle the livestats requests). This again suggests you've got different settings on your bot, which might be enough to "trip the wire" and crash the system..

I've been over the settings and over the settings, looking for what might be an error on my part... not so much concerned with differences.
The differences between the two are things that are to be expected.
Channels, etc.
And, of course, ports.

1.) Trying to narrow it down: do you think that the settings that might be different would like be settings within the .conf and/or lang file for the stats mod s ?


2.) What do you mean by "an incoming connection" though? To me, that sounds just like what it says - but it can't be a real incoming connection... it happens every time that bot is started too. What do you think it means when it appears then?

3.) I'm really wanting to pin this down, and fix it. Any suggestions on how to proceed will be appreciated.


Thanks

[nml375]

Would simply NOT starting the working bot be a good test?
In other words, start only one bot, and that bot being the one that is exhibiting the problem?

Although not conclusive, that could provide some clues whether your system crashes because your eggies are competing for some network resources. You could also try not loading the livestats-module on the working eggdrop.
You could also try copying the binaries from the working eggie to the one that does not work.

I've been over the settings and over the settings, looking for what might be an error on my part... not so much concerned with differences.
The differences between the two are things that are to be expected.
Channels, etc.
And, of course, ports.

Different port settings could be one thing, if we're dealing with a buggy network driver.
Since the crashes started when you enabled/used the livestats module, that would suggest there's something within that triggers the crash.

One thing though.. you said it only crashed on the top30 stats, not the others. This suggests this part of the module code does something different to cause the crash. Have you considered the possibility of your system not actually crashing, but simply becoming extremely sluggish from having your eggdrop stealing 99.9999% cpu-time? Try hitting Ctrl+Alt+Tab and wait for perhaps 10 minutes and see if the task manager finally appears.. you could also try using the task manager to lower your eggdrop's process' priority to make it less greedy on cpu-time.