High utilization of MongoDB (/var/lib/mongo) partition
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Released
|
High
|
Oleksiy Molchanov |
Bug Description
We faced an issue when /var/lib/mongo partition is almost full, even when ceilometer configuration is set to store meters and events only from last two days:
# grep live /etc/ceilometer
metering_
event_time_
however inside mongodb we're able to find older meters/events.
From logs which we have, it seems mongo is showing tons of errors like:
TTLMonitor] error processing ttl for db: ceilometer 0 assertion src/mongo/
I asked to provide further data and outputs from commands:
mongo
use ceilometer
show dbs
db.meter.dataSize()
db.event.dataSize()
db.resources.
db.system.
db.event.
db.event.
db.event.
db.meter.
db.stats()
db.meter.stats()
db.event.stats()
db.serverStatus()
db.getSiblingDB
db.event.
As I'm not sure if TTLMonitor is still enabled, but anyway it can suggest some strange crush of that.
I'm raising that ticket as I'm lack of other ideas which can help with further troubleshooting.
Thanks in advance for help here.
Best regards,
Jan
Changed in fuel: | |
assignee: | nobody → Oleksiy Molchanov (omolchanov) |
Changed in fuel: | |
importance: | Undecided → High |
status: | In Progress → Fix Committed |
status: | Fix Committed → In Progress |
milestone: | none → 9.2-mu-13 |
Fix proposed to branch: 9.0 /review. fuel-infra. org/41356
Change author: Roman Lubianyi <email address hidden>
Review: https:/