Redis volume usage in metrics show way higher than in instance

a month ago

I have an environment running just a test workload that has a redis instance+volume for BullMQ. In the usage metrics, the volume shows it's using almost 1GB, always growing, never decreasing. However, when I ssh into the instance and check the volume:

$ df -h /data

Filesystem      Size  Used Avail Use% Mounted on
/dev/zd34496    4.6G   28K  4.5G   1% /data

$ du -sh /data
24K     /data

Here's what the volume's Metrics tab shows:

I have two other environments, a development environment that has the same issue, and a production environment that shows ~220MB pretty consistently.

Attachments

Solved$10 Bounty

3 Replies

Status changed to Awaiting Railway Response Railway about 1 month ago


a month ago

This is a known discrepancy in which the volume metrics report higher usage than is actually on disk. It's related to how the underlying filesystem tracks space, where deleted or rewritten files can still be counted if file descriptors remain open. Redis's background save (BGSAVE) process, which forks and writes RDB snapshots, can contribute to this.


Status changed to Awaiting User Response Railway about 1 month ago


brody

This is a known discrepancy in which the volume metrics report higher usage than is actually on disk. It's related to how the underlying filesystem tracks space, where deleted or rewritten files can still be counted if file descriptors remain open. Redis's background save (BGSAVE) process, which forks and writes RDB snapshots, can contribute to this.

a month ago

How can I get true usage from the instance to count this? Or how can I clean up the open descriptors? This is the metric it’s using for billing, right?


Status changed to Awaiting Railway Response Railway 30 days ago


Status changed to Open Railway 29 days ago


a month ago

I'll just wipe my non-production redis volumes once a week or something. Very strange.


Status changed to Solved shamsup 28 days ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...