runtime error: integer divide by zero.

[jvshahid]

After creating a lot of continues queries, I keep getting lots of "divide by zero" messages.

********************************BUG********************************
Database: tracking
Query: [SELECT MEDIAN(total) as total, MEDIAN(total_kB) as total_kB, MEDIAN(uptime) as uptime, MEDIAN(reqs) as reqs, MEDIAN(bytes_per_sec) as bytes_per_sec, MEDIAN(bytes_per_req) as bytes_per_req, MEDIAN(busy_workers) as busy_workers, MEDIAN(idle_workers) as idle_workers FROM /^project.(\d+)\.apache\.16.*/ GROUP BY time(1440) where time > 1396863886702460u and time < 1396863887702444u]
Error: runtime error: integer divide by zero. Stacktrace: goroutine 18 [running]:
common.RecoverFunc(0xc222cd7aa0, 0x8, 0xc212ff5000, 0x17b, 0x0)
        /home/vagrant/influxdb/src/common/recover.go:13 +0x106
runtime.panic(0x884c40, 0x1004a9d)
        /home/vagrant/bin/go/src/pkg/runtime/panic.c:248 +0x106
cluster.(*ShardData).QueryResponseBufferSize(0xc210134460, 0xc21064eaa0, 0x64, 0x0)
        /home/vagrant/influxdb/src/cluster/shard.go:361 +0x12d
coordinator.(*CoordinatorImpl).queryShards(0xc2100c88c0, 0xc21064eaa0, 0xc2209d20f0, 0x1, 0x1, ...)
        /home/vagrant/influxdb/src/coordinator/coordinator.go:374 +0x111
coordinator.(*CoordinatorImpl).runQuerySpec(0xc2100c88c0, 0xc21064eaa0, 0x7fc57c0a7678, 0xc22dd08fc8, 0x0, ...)
        /home/vagrant/influxdb/src/coordinator/coordinator.go:414 +0x440
coordinator.(*CoordinatorImpl).runQuery(0xc2100c88c0, 0xc21cf602c0, 0x7fc57c0a7620, 0xc2100d8000, 0xc222cd7aa0, ...)
        /home/vagrant/influxdb/src/coordinator/coordinator.go:146 +0xdb
coordinator.(*CoordinatorImpl).RunQuery(0xc2100c88c0, 0x7fc57c0a7620, 0xc2100d8000, 0xc222cd7aa0, 0x8, ...)
        /home/vagrant

I suspect the query doesn't return any data, hence some calculactions afterwards seem to fail.

[pauldix]

Maybe make this a config option? Remember to have that as the default that gets set even if they don't have that option in their config file.

[jvshahid]

Hey @nichdiekuh we found the bug that you were hitting. This was caused by an assumption that the group by interval is always greater than a second. The group by interval you're using is 1440 nanoseconds (which is probably not what you wanted to do), which was truncated to 0 seconds and caused the divide by zero error. That said, as I mentioned earlier, you probably meant a day and not 1440 nanoseconds. In order to achieve this you can do group by time(1d) or group by time(24h). Read http://influxdb.org/docs/query_language/ under the Group By section for more information on group by intervals.

[nichdiekuh]

That's funny, actually that was a bug in my code as well...it was supposed to be "1400m", but I'm glad now both sides have a bug less. Thank you! :-)