Desra Blog

 i've been testing imgproxy, to handle our image serving needs, and it looks good. our existing servers are php based, and we sign and encode our urls for images.  To test out imgproxy , I wanted to simply drop it in as a replacement for our servers by sending a % of traffic.  There are many ways to do this, varnish was one, with custom code, but nginx is our go-to web server, so I had to find a way to have nginx sit in front of imgproxy and rewrite the decoded url. I settled on using njs, the cut down version of javascript that plugs into nginx as a loadable module.  Then use proxy_pass to pass the uri to javascript that will return the imgproxy compatable url, and proxy to it. a sample url would be http://foo.bar/images/c2lnbmF0dXJlZm9vaHR0cDovL3MzLWV1LXdlc3QtMS5hbWF6b25hd3MuY29tL215YnVja2V0b2ZwaG90b3MvcGhvdG9fb2ZfYV9jYXQ1fHx8MTIwMHgxMjAwfHx8fHx8fHw==.jpeg it has a sig, a bucket url, and parameters like image size. Getting nginx setup nginx.conf load_module modules/ngx_http_js

I recently had a need to work with metrics, and looking at the landscape of modern tools, went with victoriametrics. After initially using the single binary version, I went on to setup the cluster version, using two nodes for everything (free nodes in oracles free tier!) run these binaries on each node (where 10.0.2.41 and 10.0.2.40 are the addresses of the nodes) ./vmstorage-prod -retentionPeriod 5y -storageDataPath /var/lib/victoriametrics ./vminsert-prod -storageNode=10.0.2.41:8400 -storageNode=10.0.2.40:8400 -replicationFactor=2 ./vmselect-prod -storageNode=10.0.2.41:8401 -storageNode=10.0.2.40:8401 -replicationFactor=2 -dedup.minScrapeInterval=1ms front vmselect and vminsert with nginx server { listen 443 ssl; server_name metrics.foo.bar; location /insert/ { proxy_set_header Host $host; proxy_set_header X-Real-IP $remote_addr; proxy_pass http://localhost:8480; } location /select/ { proxy_set_header Host $host;

I had a need to copy data from a standalone victoriametrics system to a cluster setup. for some reason the command line wasn't intuitive, so making a note here ./vmctl vm-native --vm-native-src-addr=https://source.to.copy.from:443 --vm-native-src-user user1 --vm-native-src-password password1 --vm-native-dst-addr=https://destination.to.copy.to:443/insert/0/prometheus/api/v1/write --vm-native-filter-match='{db="db1"}'

 vmagent is part of victoriametrics, and is a lightweight prometheus scraper. create a basic prometheus.yml file that defines a host to scrape, in this example it's using SSL and basic auth global: scrape_interval: 15s evaluation_interval: 15s scrape_configs: - job_name: vmetrics static_configs: - targets: - host.to.scrape:443 scheme: https basic_auth: username: user1 password: password1 then run vmagent to use this and send to a victoriametrics node to store the metrics it scrapes ./vmagent-prod -httpListenAddr 127.0.0.1:8240 -remoteWrite.basicAuth.username user2 -remoteWrite.basicAuth.password 'password2' -remoteWrite.url=https://host.to.send.to:443/api/v1/write -promscrape.config=prometheus.dsch.yml if using the cluster version, the url in the command line might be -remoteWrite.url=https://host.to.send.to:443/insert/0/prometheus/api/v1/write

I hooked up a submeter to my fuseboard some time ago, and not too long ago I started to read it with a Pi Zero.  However every now and again I would get a crazy reading, and from logging I would see bouncing. I would see interrupts where wiringpi would report the input as 0, and then again as 0, with no change to 1 inbetween.  I read up on debouncing and how the Pi is meant to do this, and also how some people have handled it in code by removing large deviations.   In my case, the solution has been to simply add logic in an encompassing if statement, to ignore interrupts where the pin is the same value as the last interrupt, and only count interrupts where the pin is a different like this  if (digitalRead(PIN) == last_interrupt_pin) { printf("Bounce ! Last Pin = %d This Pin = %d\n", digitalRead(PIN), last_interrupt_pin); }

 I recently looked at moving from a basic electricity meter to a day/night meter, and got a little caught up in the maths, so thought I write it here. Where I live, your electricity is made up of a standing monthly charge, and a usage charge.  you by default get a meter which is an all day meter and not aware of day/night.  though smart meters are coming  (smart tarriffs some other time in the future). For free you can change your meter to be a day night meter, where the day tariff is higher than the all day tariff, but the night tariff is a lot cheaper.  The standing rate is different, but lets ignore that or now. There is a magic point where your % of electrcity used during the day vs night, will be equal to the cost of the all day tariff, and expressing this lets you figure out if you will save money moving from all day to day/night.  The maths looks like this (night rate - all day rate) / (night rate - day rate) = %day  (the percent of you 24hr usage that must be in 'day'