Help with TCP Lockup!


echo $row['what_tha'];
Today, out of the blue, a OS X box running apache serving a local online database ... was no longer reachable ... but when I ran down to the room where the system sits.... it was running... I was clicking around and everything seemed fine.. but I could not ping anything, not even localhost...

So it seems that TCP or either the interface crashed... requiring a reboot to fix ...

Does anyone know about any log_files related to interface or TCP errors or something, that would help me try and track down the cause... or know of something I could place in my crontab file to check the status and retart anything that could be restarted to fix such a TCP/Interface lock-up??



I found the following in my system.log....

5 minutes of time with an error almost every second:

Sep 28 14:52:34 localhost mach_kernel: Tx Int Timeout - Comp = 0025 Kick = 0024 Int = 01284000
Sep 28 14:52:34 localhost mach_kernel:

Sep 28 14:52:35 localhost mach_kernel: Tx Int Timeout - Comp = 0025 Kick = 0024 Int = 01284000
Sep 28 14:52:35 localhost mach_kernel:

Sep 28 14:52:36 localhost mach_kernel: Tx Int Timeout - Comp = 0025 Kick = 0024 Int = 01284000
Sep 28 14:52:36 localhost mach_kernel:

then the following at the exact time the link went down:

Sep 28 14:52:37 localhost mach_kernel: Ethernet(UniN): Link is down.
Sep 28 14:52:37 localhost mach_kernel:

The rest of the log is info after rebooting ...

What in the world is going on ?


wandering shadow
Looks like your ethernet link went down. Maybe the port got turned off? Were there any other messages about ethernet in the log?


echo $row['what_tha'];
the only entries in the system.log related to ethernet are a period of five minutes with an error every second with TX ... then the Ethernet link is down entry.... Now the system is sitting on a rack-shelf with a 3` patch cable straight to a Cisco 4400 switch... so it was not anything physical that happend ... (Well I can not think of anything other than the Cisco Switch acting up, but I never seen that happen) ....

This has gotten me worried... becuase I just removed all the NT servers a set of 3 with OS X systems to replace them and what they where doing... but what's the difference between the NT crashes that happend alot and a possible Ethernet interface problem in X ? ... I hope it was just a fluke but I want to find out and make sure ....


wandering shadow
Are you using the built in ethernet ports on the computers, or did you buy any cards to put in them? Just curious really, I haven't heard anything saying that it wouldn't work if you did.

Has this happened again? Maybe it was just a fluke.


echo $row['what_tha'];
That system has done it again .... after a short run here for a few weeks... TCP locked up yet again.... I just don't get what the problem could be ...

This worries me alot ....


wandering shadow
Hmm...2 1/2 weeks, about. Not too bad. What do you have to do to get the link back up? If it's something simple that you could do on the command line, you could write a small cron job to do it and have that job run once a week. That ought to take care of the problem, at least.


echo $row['what_tha'];
What I have to do, is "REBOOT" ... I have not found a way via cli ...

According to the system log... after a few minutes of (every second) an entry from the mach_kernel stating a failed TX on Ethernet(0) ...

An log entry appears stating Ethernet Interface has failed ... but shortly after a log entry states that the Interface is back up ... but afp and any TCP actions time out ... and you can not ping the system nor ping out from the system...

When opening Netinfo Manager ..... everything looks normal ( Link: Active .. etc....)

But, in order for things to be "Really" back up... I have to reboot