It has been reported a few times, for example: https://github.com/EOSIO/eos/issues/10690 that the trace_api_plugin performance suffers greatly from the sync() call in storing to disk. This has recently caused slow down in ci/cd as well causing tests to fail do to taking so long to write to the trace api log.
We do not sync() for the block log or SHiP logs. Seems the trade off of performance for small possibility of corrupted trace api logs is worth it.
It has been reported a few times, for example: https://github.com/EOSIO/eos/issues/10690 that the
trace_api_plugin
performance suffers greatly from thesync()
call in storing to disk. This has recently caused slow down in ci/cd as well causing tests to fail do to taking so long to write to the trace api log.We do not
sync()
for the block log or SHiP logs. Seems the trade off of performance for small possibility of corrupted trace api logs is worth it.Resolves #2331