Hi everyone, I'm running a POC on a small deployment in a kubernetes
environment and after a few minutes of load testing, the data node fails
with this message:

ss o.a.i.i.processors.cache.persistence.StorageException:* Failed to
archive WAL segment*
[srcFile=/opt/work/wal/node00-ef1e49d3-1c67-4527-9a24-bae580a5ed91/0000000000000005.wal,
dstFile=/opt/work/walarchive/node00-ef1e49d3-1c67-4527-9a24-bae580a5ed91/0000000000000065.wal.tmp]]]
org.apache.ignite.internal.processors.cache.persistence.StorageException:
Failed to archive WAL segment
[srcFile=/opt/work/wal/node00-ef1e49d3-1c67-4527-9a24-bae580a5ed91/0000000000000005.wal,
dstFile=/opt/work/walarchive/node00-ef1e49d3-1c67-4527-9a24-bae580a5ed91/0000000000000065.wal.tmp]
.....
Caused by: java.nio.file.FileSystemException:
/opt/work/wal/node00-ef1e49d3-1c67-4527-9a24-bae580a5ed91/0000000000000005.wal
->
/opt/work/walarchive/node00-ef1e49d3-1c67-4527-9a24-bae580a5ed91/0000000000000065.wal.tmp:
*No space left on device*

I have one data node, with a cache, persistence enabled and I have 3 PVC
one for each of storage, WALand WALarchive.
I load data from a kafka topic using a Kafka Streamer running in a
different pod.
Incoming load (at the topic) is about 5K records per second.
Average record size is 1.8 Kb.

Data region is configured with a maxSize of 5 Gb
Storage volumen with 10 GB
Wal volumen with 2 GB
Wal archive with 2 GB. (also tried 3 and 4)

The rest of the settings  (page size, wal segment size, etc) are with
default values.
Ignite version is 2.9.0.

My question is, Is there some recommendation on the size these volumes
should have respective on the storage size, record size or some other
factor?
Maybe wal segment? If I increase the wal segment from 64Mb (default size)
to lets say 512 Mb, How much should I increase WAL and WAL archive volumes?

Thanks,
-- 
Facundo Maldonado

Reply via email to