It pronts to log
All I’m getting is “Request scavenging because /admin/scavenge request has been received.”…
How many chunks do you have?
Do you already have one running you should get messages scavenging chunk ######
I’m not seeing those in the log… I’ll investigate whether we are doing something weird logging wise…
Doesn’t look like it’s an issue with logging… Is this what we’re seeing perhaps?
- Fixed a bug which could cause an infinite loop during scavenging when more than 10 chunks had previously been merged.
You would have to already have a scavenge running for that.
I’m not sure which version that got merged in as well but I believe 3.0.3 has the fix for it.
Even if you were hitti that it would still print that the merge had started
We restarted the server and now scavenge starts. But it doesn’t run to completion. Server runs ES 3.0.1. On a different machine with ES 3.0.3 it runs to completion on the same data. So I think we’ll upgrade…
What do you mean it doesn’t run to completion? You will always have a .000000 chunk left over (the last one), because it hasn’t ‘filled up’ yet.
If the compaction needs to compact jore than 10 files into one there was a resplved issue wherenit may jot finish
This may be affecting you.
An upgrade to 3.0.3 fixed the problem… Size of db went from 220GB to 18 GB