Hello,
I observerd some strange behaviour of Prometheus when I issued a
clean_tombstones API call.
1. WAL will not be written to disk while the compaction runs. This
means that very long running compactions can exceed the available
memory.
2. When one 'block' is copied, the old source block won't be deleted.
3. New 'blocks' won't count against storage.tsdb.retention.size and so no
old blocks are deleted until the compaction is finished.
If this is correct and metrics are removed that are stored in each and
every level then it's not advisable to clean_tombstones, as this will
require almost the same amount of disk space as the actual data?
Matthias
--
You received this message because you are subscribed to the Google Groups
"Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/prometheus-users/22562290-485d-dbc6-175a-74b1e2c53945%40zu-con.org.
