I can see the pinot-zookeeper disk usage high .. w...
# troubleshooting
s
I can see the pinot-zookeeper disk usage high .. what could be the root cause since metadata cant be 65 gb out of 95 gb given with few million records in table .. is it database indexing causing some external views to be stored on zookeeper disk? @Mayank @Xiang Fu @Jackie @Subbu Subramaniam @Kishore G
p
You need to setup a zookeeper clean up job. To remove old snapshots.
m
Thanks @Peter Pringle
s
@Xiang Fu pinot brokers again went to crashloopback status once zookeeper disk storage got full.. Do you have any input what is causing zookeeper disk space to be exhausted and how to clean it and if zookeeper is full.. Why pinot brokers crash and how to bring pinot back to running stage with all tables and data intact if such a situation happens after cleaning zookeeper Or restoring it from attached pvc on google kubernetes engine
We end up losing all the tables and data when this problem happens
Already there is a thread with description and log of the same issue which replicated again and we are getting regression of this issue

https://apache-pinot.slack.com/files/U0224CF6LMT/F02CK3E6EKG/capture3.png

x
you can config environment variables :
ZK_SNAP_RETAIN_COUNT=3
and
ZK_PURGE_INTERVAL=1
to zk statefulset
s
Ok will try that and update if issue regression happenS further
m
@beerus ^^
s
@Xiang Fu which file do we need to config these variables in helm config : 
ZK_SNAP_RETAIN_COUNT=3
  and 
ZK_PURGE_INTERVAL=1
  to zk statefulset ..
we are not aware of this cc: @Mohamed Kashifuddin @Shailesh Jha @Mohamed Hussain
x
It’s environment variables, so check values.yaml
s
do we need to add it here @Xiang Fu in pinot/values.yaml?
@Xiang Fu I am worried if setting zk snap retention period in ZK might have some consequence. AS it store Metadata.  about all the tables and data. will there be some consequences?(means attaching the pvc back when restarting pinot pods after helm upgrade .. will tables data remain intact?) Because If We set retention as 3 day and metadata cleaned. What about older meta data / zk znodes?? can you enlighten a bit
or need to add here ?
x
Yes
s
x
this retention is for zk snapshots, your data are still full data.
🍷 1
s
Thanks @Xiang Fu
Hi @Xiang Fu..even after adding these environment variables and updating helm ..the disk usage keeps on increasing .. do i also need to do kubectl delete po dev-pinot-zookeeper-0 dev-pinot-zookeeper-1 dev-pinot-zookeeper-2 -n dev-pinot in order to get the environment variable changes to actually work for setting up cron job to clean up old snapshots and only helm upgrade is not enough ?
x
then it means you are not setting it correctly, please read the script to start zookeeper and figure out the environment variable name
s
x
also enter into the k8s pod to check the start script to see how those configs are set
s
sure @Xiang Fu .. will check and update
l
hey Sadim did you figure how to fix this? We are seeing kinda the same issue with our zookeeper images going up in size
s
Enabling autopurge in zookeeper config should resolve this issue @Luis Fernandez
👍 2