Freezing Machine?
 

I went an installed slackware 10.1, updated & recompiled the kernel to 2.6.10. The machine is used exclusively for web development, apache, php4 and mysql are installed. only mysql is from the slackware packages, php & apache are installed from source.

SSH is listening for connections, although it will only allow authorised users with keys (no passwords), also CVS is running, but again, only keyfile and user restricted access.

The problem is that I'm having problems with seemingly random lock ups. Quite what exactly is happening I am not sure because I am currently away from the machine itself (but I can get someone to restart the machine for me every so often).

The first time this happened, it was a kernel panic, saying interrupt exception (or something similar), which I found in the forums to be pointing at IRQ sharing conflicts. I then had a soundblaster live and tekram SCSI controller removed from the machine. Also I went through the bios and disabled nearly everything that would take up any IRQ resources. Only a network card and graphics card are left in the machine. However, no matter how I try, it seems USB is intent on sharing an IRQ with my network card, I've even tried moving the card into a different PCI slot.

I'm thinking it might be a power (ACPI or APM) related issue. In /etc/rc.d/rc.M , I have changed it so that it just says setterm -blank 0, instead of what's already there with powerdown, powersave, etc. Can someone provide a good guide about how to get proper power up and running in slackware? I think on the default install of slack, I enabled modprobe apmd to enable shutdown of the computer when you issue 'shutdown -h now' (otherwise it just halts and that's it)

Now, the machine hangs after around 5/6 hours uptime, and it just sits there with a blank screen I think. (I've tried moving the mouse and it doesn't display again, also the machine stops responding to pings, can't SSH in, etc).

After a crash, I try get into the machine and check everywhere under /var/log but there seems to be no trace of there being a system crash. What's the best place to look after a crash?

I'm sorry for so many questions, but I am a relative newbie to the world of Linux (Slackware) and I'm learning all the time. I'm at a bit of a loss now with what to do.

If it helps, my system specs are:

Intel Pentium III 450
768Mb RAM
Abit BX6 Motherboard
GeForce 1 32Mb DDR
Generic SIS900 network card
Hard drive

I also have to say that I ran memtest86 to check the ram of the computer (shouldn't really be a problem) and that came back with no errors. While I don't know of any stress/stability testing programs for linux, a kernel make and compile went through without problems, so this probably suggests a kernal problem?

I had a very similar problem after upgrading to 2.6.10 in the /testing directory. I never did find the exact cause, but the remedy was to ignore the 2.6.10 upgrade instructions and compile a new kernel from the source using a modified .config from 2.4.29.

This removed all my problems and have been stable for 2-3 months.

samac

Oh yeah I forgot to mention that I have samba running on the machine to connect with some other windows clients on an internal lan. While I was looking through some of the logs for reasons why the machine is hanging, I noticed a lot of messages from smbd, most notably saying, invalid client 0.0.0.0 ? Anyone have any idea about this, a search on the forum seemed to return nothing..

And, J.I.C., let's not forget to get a good ram-checker & run that to see if you have a hardware problem...

I don't really know why, but my instincts are telling me to write this, and since they are right *far* more often than not, get & use Memtest86 or the program (dosfloppy image) from www.simtester.com. [ double 'm's?] Simtester saved my butt once, so do this, J.I.C.
LOL

Hi,

I have already run memtest86 and that has returned no errors (ran for 6 hours straight).

After the latest crash, it just hung on the console screen asking for login information. Completely frozen, no response. Checked in /var/log as always to find nothing useful about said crash.

Anyway, I have decided to recompile kernel to the latest 2.6.11.5. It's been up now for about 3 hours, let's hope that it's fixed whatever it was that's going wrong.

If it still keeps crashing, then I think that I'm going to have to revert back to a default kernel that was shipped with Slackware 10.1

Thank you for your help so far, any more would be appreciated!

Maybe you found the answer.
samac said start from [older] config. Is that what you did?
If so, maybe you'll get the same results -- 2 months+ of Ok.

Samba: I don't know this one yet, but 0.0.0.0 ?? this is a totally invalid address. Did logs show anything about what samba was doing when this happened?
If so *AND* if you like, post part of log or zip-&-email to me.

Actually, I just recompiled from scratch as I heard that there are some options from the old kernel (2.4.x) that are incompatible with 2.6.x kernels.

The machine has been up for about 9 hours now, so I suppose that is a good sign.

Regarding the problem with the strange ip address 0.0.0.0, I have found these messages in:

/var/log/messages:
Mar 22 19:21:51 machinename sshd[2146]: Server listening on 0.0.0.0 port 22.

/var/log/syslog
Mar 22 19:07:06 machinename smbd[2668]: Denied connection from (0.0.0.0)
Mar 22 19:07:06 machinename smbd[2668]: Connection denied from 0.0.0.0

Also, there is a samba.0.0.0.0 error log file (it has one for every other client ip address too)

I do have ftp up on this machine, but it is pretty locked down (only two user accounts are allowed to log in and are chrooted, no root or other logins allowed). I noticed two attempted logins from seperate ip addresses, but I guess that they must have realised it's quite difficult to get in.

Thanks for your help.

Samba I cannot help you with; I don't know it yet.
It does look possible that somehow the 0.0.0.0 is from inside your machine itself, thus probably not a security issue. I said -probably-...

As for no root and chrooted: well done! :)

Best,