0

I am using a server at OVH. And I never had any problems until now.

Apparently the server seems to work without any problems, I mean no HW faults. We have run many HW tests for RAM, CPU and disks. Nothing pop up. It means that for the moment the HW seems out of the equation.

According to OVH they don't reboot the server, it is the OS that does such action when the CPU and RAM is used at 100%.

We are using Debian 9 at the latest level of packages.

What happens is that we use that server for performance tests, and also heavy loads testing. It means that we are very often using the server at the limit of what he can deliver. So having a usage of CPU with 100% or RAM also at 100% is not rare.

What is strange is that now, every time such huge load is created, the server is restarted. We didn't have such behavior in the past with older version of Linux (Debian 6/7, Ubuntu 16.04, Centos 6/7).

From the logs we see nothing, it is like nothing is written, until the server is (automatically) rebooted.

I have googled to understand what could in Linux create such behavior, and found some kernel parameters like

kernel.panic

found some info here: https://unix.stackexchange.com/questions/29567/how-to-configure-the-linux-kernel-to-reboot-on-panic

But on my system the params are set to 0 -> meaning wait for ever. No reboot setup.

Is there any other setting to prevent the server to reboot without any warnings?

Thank you in advance

Alessandro Perucchi
  • 61
  • 1
  • 8
  • did you disabled the OVH monitoring? check on ovh server panel, if you find some log – c4f4t0r Jun 18 '18 at 15:33
  • I could see 2 things: Either a thermal problem coming up under 100% load, or some watchdog doing a reset because it suspects a runaway process. Your image of OVH? In the later case, their support should know. – TomTom Jun 18 '18 at 15:45
  • 100% memory usage can make the system unstable. Normally in such a situation, the OOM Killer tries to free up memory by killing some process. Please provide the settings of `/proc/sys/vm/oom_kill_allocating_task`, `/proc/sys/vm/panic_on_oom`, `/proc/sys/vm/overcommit_memory`, `/proc/sys/vm/overcommit_ratio`, `/proc/sys/vm/overcommit_kbytes` and `/proc/sys/vm/oom_dump_tasks`. See https://www.kernel.org/doc/Documentation/sysctl/vm.txt – rda Jun 18 '18 at 17:47

0 Answers0