[LU-19043] Check vm.dirty_ratio at client mount time

• Type: Improvement
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.17.0
• Severity: 3
• Assignee: Patrick Farrell
• Reporter: Patrick Farrell

Description

Lustre's writeback support does not work well with vm_dirty_ratio set to 0.  There is ongoing work to resolve this in LU-19014, but in the meantime, we should issue a warning at mount time.

To prevent potential write performance degradation, the mount utility should verify that the vm.dirty_ratio sysctl setting is configured to a value of 10 or greater. Values below this threshold can cause inefficient write behavior under load. If the value is found to be too low, the utility should issue a warning and suggest the following fix:

 Consider increasing vm.dirty_ratio to at least 10 using:
sysctl -w vm.dirty_ratio=10

Sample code to check the value in C:

int dirty_ratio = 0;
FILE *fp = fopen("/proc/sys/vm/dirty_ratio", "r");
if (fp) {
    if (fscanf(fp, "%d", &dirty_ratio) == 1 && dirty_ratio < 10) {
        fprintf(stderr, "Warning: vm.dirty_ratio is set to %d, which may hurt performance\n", dirty_ratio);
        fprintf(stderr, "Consider increasing it to at least 10 using:\n");
        fprintf(stderr, "  sysctl -w vm.dirty_ratio=10\n");
    }
    fclose(fp);
} 

This check should be performed during the mount process to alert users before the file system is fully mounted.

Issue Links

• is related to: LU-19014 Random client hung in balance_dirty_pages() with CGroup (memcg) enabled