Also check dmesg for any weird messages. I’ve experienced weird issues at work where one CPU core will softlock in kernel code (watchdog: BUG: soft lockup - CPU#21 stuck for 23s!), and whatever was supposed to be running on that core will just… freeze for a while. We have a kernel team at work and they haven’t been able to track down the problem yet.
I didn’t actually know about THP until now… Thanks for the info @WSS and @FHR.
Me neither. There’s not much of such issues in dmesg but I did find this:
[ 0.700275] pmd_set_huge: Cannot satisfy [mem 0x80000000-0x80200000] with a huge-page mapping due to MTRR override.
Have to read more.
The preferred method to disable Transparent HugePages is to add “transparent_hugepage=never” to the kernel boot line in the “/etc/grub.conf” file. The server must be rebooted for this to take effect. Alternatively you can add the command to your /etc/rc.local file.
If you’re using this information for your current employer, my fee is $5,000/hr. I spent about a week on this.
Yeah that can also be caused by THP actually. Of course RedHat knows about this issue:
Possible, but when it’s always the same CPU, it’s probably not.
Hey,
Can I buy y’all a coffee or beer please?
I would like to request that you donate whatever you feel assistance was worth to Miguels’ rescue fund, please. Obviously, I speak only for myself.
Yup, had already commented on that post before posting here. I haven’t read the full post so wasn’t aware if funds are still needed therefore asked before.