I have three FreeBSD boxs and all of them freez after 26 hours uptime

mfaridi

mfaridi

I use FreeBSD for NAT and samba server ,before FreeBSD 8 and 8.2 , I use FreeBSD 6 and it was good for me and I run many services with it like NAT server with PF and file sharing server with samba and I have many good uptimes with FreeBSD 6 like 190 days and FreeBSD work for me like charm
after relaese FreeBSD 8.2 , I have clean install of FreeBSD 8.2 AMD 64 and start make NAT server and Samba server with it , but after two days , I want SSH to my server and I want check activity of server , and I see I can not ssh to server and after that I go to my work place and I see my FreeBSD system is hang and Freez and I reboot my FreeBSD box with reset button of case , after reset my system start working good , but after two days , I see my system hang again and I have to hard reset it , so I attach mouse and keyboard and monitor on my FreeBSD box and start control what happen in my FreeBSD box and I see my system Freez and agian after 26 hours uptime and I see error about kernel panic on monitor , and again I hard reset system and after 26 hours , I see error about kernel panic on monitor and my system hange and freez again .
after one week , I think this is hardware problem and I buy another system with new mainboard and new CPU and new HDD and RAM and install on it FreeBSSD 8.2 AMD 64 , but after 26 hours my system hang again and I see error about kernel panic , and I have to hard reset system again .
I install FreeBSD 8.2 AMD 64 on another system and I see Freez and hang again.
I check these in three systems
my first system is HP server with xeon CPU and two SCSI HDD
my next system is gigabyte mainboard with AMD CPU and two SATA HDD and Nvidia card with 4GB RAM
my next system is again gigabyte mainboard with on board VGA and 2 GB of ram and one HDD.

For NAT server , I use PF firewall and for enable PF and ALTQ options , I use GENERIC kernel , and I put that options in Gereric kernel and compile kerenl ,

my qustions are these
1- why my FreeBSD box Freez and hang after 26 Hours and I have to hard reset all of them ?

2- where I can find error log when my FreeBSD box start Freez and hang

I do not have this problem with FreeBSD 6 and I do not check FreeBSD 7
 
I don't have enough information to help you, but it happened me once with a machine that, even if working and behavir right, after a few days was unresponsive. It appeared in the network, it was pong-ing in reply to ping, but I was unable to SSH in and to log in from the console. All the machine was locked. It turned out it was an electrical problem, and placing the machine behind a power stabilizer solved the problem. However I was unable to see the problem from the logs.
Another thing I would check is the temperature of the machine, even if this should appear in the logs.
 
we have power stabilizer and other system works good , we have three FreeBSD systems , 24 Windows system and one Linux system . all of them work good without problems , but FreeBSD box hang and Freez after after 26 hours
 
Do they freeze after 26 hours or at a specific time?
So, if you would reboot one at 12:00, one at 14:00 and one at 16:00, do they freeze at 14:00/16:00/18:00 or all at the same time?
What time of day would that be, then? Noon would point towards either heat or electric surges from microwave ovens, f.e. Speaking of microwaves, is the surounding area clean of equipment which can cause EM waves? (We had crashes of systems in a company I was working at because someone used a bad electric welding machine the floor below us. Since he moved about, the crashes followed him. That made for some interesting tracking.)

Did you add anything to cron?
 
my system freez after 26 hours ,after last reboot , for example when I reboot it at 7 oclock , it freez after 26 hours , and when I reboot it at 14 oclock it freez after 26 hours , I do not put anything in crontab
 
The 3 systems are pretty different, and yet they freeze after the same amount of time. So where is something common between them? To me, this starts to smell strongly of memory leaking in the kernel somehow. The KVA_PAGEs are the same in all 3 systems, regardless of system use and memory size.
What configuration settings are common in all of your systems? do they all, for example, use the same type of network card?
What would help in debugging would be getting the memory status (maybe via sysctrl) and mail it to you every 2 hours. If the kernel memory grows and grows, you can then verify if it would hit the maximum after 26 hours.
 
Check your memory usage. There is a problem with 8.2-Release when used together with Samba, memory will just disappear. The fix is to upgrade to 8-Stable
 
olav said:
Check your memory usage. There is a problem with 8.2-Release when used together with Samba, memory will just disappear. The fix is to upgrade to 8-Stable
Click to expand...
Ah!!! This is serious, I didn't had this problem on physical machines (and now I admin like 8 phisycal FreeBSD 8.2-Release machine), then I had it on a virtual FreeBSD 8.2-Release installation under VMware Workstation 8. I was blaming the new VMware version, then the problem is exactly the same as described here. After, in my case, 22 hours, the system freeze. And I do have to reset it exactly like if it was a real machine.

This machine is a Internet gateway for a virtual machines infrastructure running firewall and forwarding.

For now I replaced it with a Linux Debian 6 machine until I do not understand what happen under the hood. BTW, I also do not have too much time to lose with it :)

Do someone in the developers team is looking into this really annoying and SERIOUS problem?
 
Yes, the problem has been fixed for many months already. The thing is, bugfixes doesn't make it to RELEASE branch until the next version. In this case it's the 8.3-RELEASE. If you want the fix now, you have to upgrade to the STABLE branch. Or you can apply the patch and recompile the kernel.

I find the STABLE branch a much better branch than the RELEASE branch, so I'm a STABLE fanboy now :e
 
You must log in or register to reply here.
Back
Top