How can I resolve this error: KVStorageProvider --- saveBatchData:upsert --- No collection available

[moorvogi]

i keep seeing this error in the internal logs

kvstorageprovider - an error occurred during the last operation ('savebatchdata:upsert', domain: '0', code: '0'): No collection available.

we aren't using KVStores, i've enve gone in and disabled the kvstore via web.conf. it's still kicking this error. i found a similar post about empty bulk data and tried the recommended query against splunkd_access sourcetype. i have ZERO with code 500 as the other post indicates to check for.

also, no servers in the cluster have the KVStore role.

[moorvogi]

It was still kicking the error because of an error on my part. It's no longer throwing the error because the web.conf is now set properly. apparently spelling matters... hahaha

but, in the event we want to use KVStores later, i'm still going to try to find the answer.

[vr2312]

I reckon you would want to sort out the kvstore errors first, one search head at a time and then observe if this is still being observed.

[vr2312]

Hello @moorvogi

You might want to resync/clean the KV Store by performing the following steps :

1. Stop the Search Head
2. Run this command locally : splunk clean kvstore --local
3. Start Search Head

[benedict]

hello @vr2312  i am curious if i ran the splunk clen kvstore --local  commands will it erase the data in the storageEngine?                                      

You might want to resync/clean the KV Store by performing the following steps :

1. Stop the Search Head
2. Run this command locally : splunk clean kvstore --local
3. Start Search Head

[moorvogi]

top of the mornin to ya.

rgr on process and steps. i haven't tried steps in that order. to clarify, do i need to stop ALL search heads then do the command on each one before restarting each search head? This error was bouncing around between them. Seemingly "at random", though i'm sure there's a pattern somewhere.

[vr2312]

You would want to use the clean command on the SH where you are observing the error. There is no need to stop all search heads to perform this on one particular Search Head.