When I start up influxdb, I try to load all my continuous queries in batches (in order to downsample the many metrics we have).
Originally I tried to POST one query with about 1,000 CREATE CONTINUOUS QUERY
statements, and that would fail, so I started sending them in batches of 250, but still occasionally see a failure.
The failures do not happen at a consistent query number, and seem to be some back-pressure mechanism.
Q: Are there any best practices for loading many Continuous Queries?
Q: Is this a bug that influx doesn’t successfully handle all of the queries?
Here’s some of the influx responses from my latest run:
Batch 1 (success)
{
"results" : [
{
"statement_id" : 0
},
{
"statement_id" : 1
},
...
{
"statement_id" : 248
},
{
"statement_id" : 249
}
]
}
Batch 2 (failure)
{
"results" : [
{
"statement_id" : 0
},
{
"statement_id" : 1
},
...
{
"statement_id" : 220
},
{
"statement_id" : 221
},
{
"error" : "not executed",
"statement_id" : 222
}
]
}