Strange error

[fixx]

Hi Guys

I have a CentOS 5.8 box running kernel 2.6.18-308.20.1.el5PAE in a data center in Bryanston. Last week they moved the machine to another rack and after that, every 48 hours the machine stops accepting TCP connections. It still responds to pings, and has the following kernel error on the terminal screen:

host byte =did_bad_target driverbyte=driver_ok

So after some searching around some people reckon its a failing hard drive and other reckon its an ACPI issue.

I do have smartd monitoring my hard disks (2 x SATA 2TB Seagate drives in a md raid 1 array), which reports no errors on the drives.

I have the following boot options added yo my kernel:

kernel /vmlinuz-2.6.18-308.20.1.el5PAE ro root=/dev/md2 acpi=off acpi=no noacpi noapic apm=off

I am currently running a badblock scan on the drives but so far nothing.

Any ideas?

 

[Nuro]

Hi Guys

I have a CentOS 5.8 box running kernel 2.6.18-308.20.1.el5PAE in a data center in Bryanston. Last week they moved the machine to another rack and after that, every 48 hours the machine stops accepting TCP connections. It still responds to pings, and has the following kernel error on the terminal screen:



So after some searching around some people reckon its a failing hard drive and other reckon its an ACPI issue.

I do have smartd monitoring my hard disks (2 x SATA 2TB Seagate drives in a md raid 1 array), which reports no errors on the drives.

I have the following boot options added yo my kernel:



I am currently running a badblock scan on the drives but so far nothing.

Any ideas?

I'm not sure if the error is related. Just a couple of questions:

1. Are you sure it's your machine responding to the pings? (happened to me once)
2. Any interface errors in ifconfig?
3. Does syslog have anything interesting at the time you lose connectivity, or perhaps dmesg.
4. Do all services die? (e.g. ssh/ftp/ntp)

The last time I had this was with a bad realtek driver, but that was years ago.