victorialogs crash about victoriametrics

[valyala]

@vZ2JEZsZW , thanks for filing the bugreport!

It looks like VictoriaLogs contains only 56 log entries according to the following line from VictoriaLogs log:

successfully opened storage in 0.005 seconds; smallParts: 7; bigParts: 0; smallPartBlocks: 7; bigPartBlocks: 0; smallPartRows: 56; bigPartRows: 0; smallPartSize: 10090 bytes; bigPartSize: 0 bytes

@vZ2JEZsZW , could you provide all these log entries in JSON line form? This will allow reproducing the issue and fixing it faster. The ingested log entries can be collected from the logs generated by VictoriaLogs if it is started with -logIngestedRows command-line flag as described here. Another option is to pass debug=1 query arg to HTTP-based data ingestion APIs. You can pass debug=1 arg to HTTP-based VictoriaLogs data ingestion API at log shipper side. Note that VictoriaLogs doesn't store the ingested logs when they are written to VictoriaLogs log output.

[lujiajing1126]

Same issue

This leads to vl crash

[vZ2JEZsZW]

-logIngestedRows

Tried running it with -logIngestedRows, but there was no rows on the container output. Apparently this requires sending the new records, but I can't do the ingestion currently. (not actually using it, but watching occasionally if I can use it). Also, I assume sending new records would change the state of data files. Attached is data volume in its current state issue6258-viclogsdatavol.tar.gz just in case this can be of any help for debugging

[valyala]

@vZ2JEZsZW , thanks for the provided data! The issue should be fixed in VictoriaLogs v0.6.1.