Linux: q3ded Signal 11 on map load

[NjR_Shrapnel]

Hey, all!

Wow. Good to see these forums back. I've been lurking a bit, trying to remember my old password so I could use my old account, but alas....

Anyway, I'm hoping some linux guru out there can help me.

I run a couple q3 linux servers using q3ded. One machine works great, but because of the heavy load I want to move a q3 server to another, almost identical, machine. Both run Fedora Core 3. On the new machine, I can run q3ded and it's fine until the map load. Then, I get this...
--------------------
32736 files in pk3 files
Loading vm file vm/qagame.qvm.
VM file qagame compiled to 1951934 bytes of code
qagame loaded in 5055200 bytes on the hunk
Received signal 11, exiting...
Shutdown tty console
--------------------
Q3 file structure and configs and start commands are exactly the same on both machines, so I'm thinking something in the linux memory setup must be bad. But I don't know much about that.

Server hosting company has done a complete check of hardware - ram, motherboard, etc. All looks good. And since it's q3ded, I assume the problem has nothing to do with the display or video drivers...

Only difference between the two machines is one is a 2.4Ghz Celeron and the other is a 2.2Ghz. And, obviously, there must be some OS config differences even though both are Fedora Core 3. Re-installing the OS or cloning the drive isn't really an option.

Any ideas would be greatly appreciated!

Thanks in advance.

- NjR_Shrapnel

[AmIdYfReAk]

what map are you loading?

have you tryed to lead a diffrent map?

[^misantropia^]

Signal 11 means that something is causing a segmentation fault. Any particular mod you're running?

[NjR_Shrapnel]

Thanks for the replies...

1) I've tried a number of maps. CTF and DM. Always signal 11.
2) Running just the basic q3 - no mods.

I've tried a variety of com_hunkmegs settings, too. Only thing that doesn't generate a signal 11 is if I set com_hunkmegs so low it won't load the map.

Thanks again,

- NjR_Shrapnel

[^misantropia^]

What are the md5sums of the pak files and the quake3 binary?

[^misantropia^]

This is what the sums for PR 1.32 should be, btw:

Code: Select all

1197ca3df1e65f3c380f8abc10ca43bf  baseq3/pak0.pk3
48911719d91be25adb957f2d325db4a0  baseq3/pak1.pk3
d550ce896130c47166ca44b53f8a670a  baseq3/pak2.pk3
968dfd0f30dad67056115c8e92344ddc  baseq3/pak3.pk3
24bb1f4fcabd95f6e320c0e2f62f19ca  baseq3/pak4.pk3
734dcd06d2cbc7a16432ff6697f1c5ba  baseq3/pak5.pk3
873888a73055c023f6c38b8ca3f2ce05  baseq3/pak6.pk3
8fd38c53ed814b64f6ab03b5290965e4  baseq3/pak7.pk3
d8b96d429ca4a9c289071cb7e77e14d2  baseq3/pak8.pk3
3581d6ad496ebb7f922adc36451cd708  q3ded

[NjR_Shrapnel]

Interesting... all the pk3s check out, but the q3ded is...

f71a7ee9d8653b53cd56076a9120f7dd

I checked the other server (that runs fine) and the q3ded md5sum is the same on that one as above.

I am running 1.32.

[AmIdYfReAk]

i know this sounds stupid, but have you tryed copying the md5ded from the running box? or maby downloading another one?

[NjR_Shrapnel]

Thanks.

Just tried that. Same problem.

I'm pretty sure it's something in the linux virtual memory setup. I'll keep trying and am open for any other suggestions...

- Shrapnel

[^misantropia^]

Which version of 1.32 are you running? For Linux, the latest official PR is 1.32b-2.

[NjR_Shrapnel]

Thanks again!

But, unfortunately, it still gives me signal 11.

Actually, the version I got from id's FTP site is.

linuxq3apoint-1.32b-3.x86.run

I did find out that the server that runs everything fine is Fedora Core 3 (2.6.9-1.681_FC3) and the one that's giving me troubles is - Fedora Core 3 (2.6.12-1.1378_FC3). Looks like the problem one is newer....

Swapfiles seem to be identical on both machines...

- Shrapnel