508 virtual memory errors, but no memory limits
We updated a server to CL kernel version 3.10.0-614.10.2.lve1.4.50.el7.x86_64 yesterday. We also gave in to the Security Advisor nags and had Kernelcare installed at the same time.
This morning we noticed that all WordPress websites on the server are now giving 508 errors. We have been in touch with cPanel support (ticket 8521131). So far without much success, except that two different supporters so far have told us that we need to raise the memory limits for our users.
The point is, though, that we have NOT set any memory limits at all - these accounts are all set to unlimited.
Anyone else seeing the same?
I suspect that the "0" that is supposed to mean "unlimited" is in fact being interpreted as "0" memory available to our users. (We haven't tried working around it by setting an arbitrary limit yet, ticket is still open).
/ Tom
-
Hi Tom, I've not seen any similar reports. Your support ticket has been escalated to CloudLinux so they can take a closer look and determine if this is a bug. I'll monitor the ticket and update this thread with the outcome. Thank you. 0 -
Nothing from cPanel or CloudLinux support since the "escalation". We cannot keep waiting, so we've temporarily reverted to kernel 3.10.0-427.36.1.lve1.4.47.el7.x86_64. That kernel works fine. 0 -
I also had an issue with Apache (EA4) not loading any sites when booting into kernel 3.10.0-614.10.2.lve1.4.50.el7.x86_64. I downgraded back to the previous production kernel to resolve it. 0 -
The issue is happening only when pmem is set to 0 (unlimited), temporary workaround is to make it 1G for example. 0 -
Well, we tried setting hard limits, but that just gave us 503 errors instead (we are using Litespeed). 0 -
Hello, To update, a bug report is open with CloudLinux to address this issue. The case number is CLKRN-124. I'll update this thread with more information on the status of this case as it becomes available. Thank you. 0 -
This has been already fixed in latest beta kernel: Beta: CloudLinux 7 kernel updated 0 -
Hello, The previous post is correct. This is addressed as part kernel version 3.10.0-614.10.2.lve1.4.51: Beta: CloudLinux 7 kernel updated You can follow the CloudLinux blog to see when it's released as a stable version: CloudLinux - Main | New template Thank you. 0 -
Actually, no, it isn't fixed. Latest kernel that works for us is 3.10.0-427.36.1.lve1.4.47.el7.x86_64. The most recent 3.10.0-614.10.2.lve1.4.55.el7.x86_64 kernel gives the same 508 error. 0 -
Actually, no, it isn't fixed. Latest kernel that works for us is 3.10.0-427.36.1.lve1.4.47.el7.x86_64. The most recent 3.10.0-614.10.2.lve1.4.55.el7.x86_64 kernel gives the same 508 error.
Hello Tom, I'm not seeing a reference to the case that addresses the issue in any of the production versions of the CloudLinux kernel: CLKRN-124: fixed memory allocation error in container due to overflow in mem_cgroup_apply_beancounter; It's still only released to the beta version of the CloudLinux kernel from what I see. The most recent production release of the kernel only shows a CVE fix in the change log. Thank you.0
Please sign in to leave a comment.
Comments
10 comments