wejick/alive

Periodically test endpoint in location and ISP aware

alive's Issues

Create Dockerfile for alive-server

Edge computing?

So the idea is not to have 1 alive-server running on the cloud, but instead to have alive-server deployed near the agents. This will complicate and make the deployment more expensive. For what benefit?

Add flag option for config

so we can do
alive-server --config=config-file.json --db=alive.db

Invalid timestamp when running fluentbit on podman

I wanted to make development, testing and deployment easier by running fluentbit on podman, however I keep getting issue.

Running the binary build from source is fine, however running it on top of podman product this error :

1970-01-01T00:00:00.000000000Z API_latency_histo{ISP="Indiehome",domain="google.com",geohash="qqguzgberuhd1",http_code="200 OK",location="Jakarta",method="GET",name="Get Google.com",test_result="PASS"} = { buckets = { 0.2=7, 1=70, 5=70, 25=70, 125=70, +Inf=70 }, sum=23.3202, count=70 }
1970-01-01T00:00:00.000000000Z go_gc_heap_allocs_by_size_bytes_total = { buckets = { 9=8862, 25=76814, 65=159799, 145=200524, 321=209921, 705=214142, 1537=217015, 3201=217921, 6529=218752, 13569=219102, 27265=219231, +Inf=219264 }, sum=2.92653e+07, count=219264 }
1970-01-01T00:00:00.000000000Z go_gc_heap_frees_by_size_bytes_total = { buckets = { 9=4743, 25=64717, 65=140968, 145=177728, 321=185388, 705=188697, 1537=191188, 3201=191967, 6529=192718, 13569=193012, 27265=193138, +Inf=193170 }, sum=2.58607e+07, count=193170 }
[2022/10/24 15:38:06] [error] [output:prometheus_remote_write:prometheus_remote_write.1] prometheus-prod-10-prod-us-central-0.grafana.net:443, HTTP status=400
user=345570: the sample has been rejected because its timestamp is too old (err-mimir-sample-timestamp-too-old). The affected sample has timestamp 1970-01-01T00:00:00Z and is from series {ISP="Indiehome", __name__="API_latency_histo_count", domain="google.com", geohash="qqguzgberuhd1", http_code="200 OK", location="Jakarta", method="GET", name="Get Google.com", test_result="PASS"}

I have passed the -tz=local to adjust the timezone, but it doesn't work. Until this one got fixed, I'll need to work with manually build or distro build.

Create Dockerfile for alive-client

Should we use mqtt instead for metric push?

The current design is to scrape the data from the agent using fluentbit. However, this might be a problem when a network partition happens and the data is not properly scrapped.

How do we use MQTT?
The naive way is to push every event to MQTT and then process it on the server, this way is very inefficient.
Another way to do it is to push the aggregated metric similar to Prometheus /metric, if fluentbit can consume from MQTT in this format, that will be really great. Or may be we can explore fluentd if that's not possible.