error: time outside range when using udp transport and influxdb is configured to microseconds

[glitterkitty]

when points get written to influxdb using udp-transport, the lib is hardcoded to use nanosecond timestamps here:

LineProtocol.cxx
146:    return line.append(" ")
147:        .append(std::to_string(std::chrono::duration_cast<std::chrono::nanoseconds>(point.getTimestamp().time_since_epoch()).count()));

in case influxdb is configured to use e.g. microseconds timestamps in influxdb.conf like so:

[udp]
  # InfluxDB precision for timestamps on received points ("" or "n", "u", "ms", "s", "m", "h")
  precision = "u"

, an error is showing in the influxdb logs:

influxd[5104]: ts=2023-03-28T22:39:51.380112Z lvl=info msg="Failed to parse points" log_id=0gras_r0000 service=udp error="unable to parse 'measure t4=13.369999999999999218 1680043191378739991': time outside range -9223372036854775806 - 9223372036854775806"

workaround:

• do not configure influxdb to anything else than nanoseconds timestamps

expected behaviour:

• points do get written to db

actual behaviour:

• points do not get written to db

Used version, compiler and platform:

• influxdb version: 1.6.7~rc0-1ubuntu0.22.04.1
• compiler: gcc version 11.3.0 (Ubuntu 11.3.0-1ubuntu1~22.04)
• platform:
  uname -a: Linux hostname 5.15.0-60-generic 66-Ubuntu SMP Fri Jan 20 14:29:49 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux
  distro: KDE neon 5.26, Release: 22.04, Codename: jammy

example / steps to reproduce:

    
const std::string db_url = "udp://localhost:8089";
auto influxdb_conn = influxdb::InfluxDBFactory::Get(db_url);

try {

    // pump in some values
    for (int var = 0; var < 200; ++var) {
        influxdb_conn->write(influxdb::Point{"measure"}.addField("t4", 73.37) );
        usleep(100);
    }

}  catch(std::exception& e) {
    std::cout << std::endl <<  e.what();
}
  

[offa]

Thanks for the report. This is similar to #86. I'm going to plan this for the next release.