sragraha@redhat.com has uploaded this change for review.
Fix: skip Prometheus inode metrics when FSAL reports unsupported values (CephFS)
CephFS returns -1 for statvfs inode fields (f_files/f_favail), causing
Ganesha to publish export_available_file_count=-1 and export_inode_utilization=100%
— triggering false inode-full alerts. This fix validates inode fields before
publishing, omitting inode gauges when unsupported, while keeping byte metrics
unaffected. Also fixes the misleading HELP text on export_available_file_count
and adds debug logging when inode stats are unavailable.
Change-Id: I53f6382fe559439233a2739f9a8ec6b0b9cf8f6e
Signed-off-by: Sreedhar Agraharam <sragraha@redhat.com>
---
M src/monitoring/dynamic_metrics.cc
M src/monitoring/prometheus_exposer.cc
2 files changed, 33 insertions(+), 11 deletions(-)
git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/41/1254141/1
To view, visit change 1254141. To unsubscribe, or for help writing mail filters, visit settings.