text stringlengths 9 408 |
|---|
[1336.52 → 1337.20] Yeah, we did, actually. |
[1337.48 → 1342.58] And I was quite surprised to see just how many people either already used Prometheus |
[1342.58 → 1347.62] and wanted to tell me all about how awesome it was or didn't use it and wanted to know more. |
[1347.70 → 1348.76] So here's a little primer. |
[1349.46 → 1355.30] Essentially, Prometheus scrapes data on remote servers, gathers it into one place. |
[1355.30 → 1364.74] You write queries to query that data and present it in a dashboard or forward on certain things to something called Alert Manager, |
[1365.02 → 1366.86] which can then alert you. |
[1367.00 → 1371.76] So if your disk percentage, for example, on a remote system goes above 80% used, |
[1372.26 → 1374.62] you could have it notify you of that and say, |
[1374.98 → 1378.64] hey, you should probably go and delete some stuff before this server fills up. |
[1378.64 → 1383.98] Now, what I ended up using it for is my VMware box downstairs. |
[1385.14 → 1386.50] It's now just my Home Lab. |
[1386.60 → 1387.80] It's not actually my main server. |
[1387.92 → 1391.28] It's just performing OpenShift Home Lab duties for me. |
[1391.86 → 1396.12] And I put it into a case where fan control was a bit of an issue. |
[1396.42 → 1400.84] And so I mentioned in the previous episode about a fan controller, |
[1400.84 → 1406.56] the Corsair Commander Pro that is usable on Linux now. |
[1406.80 → 1415.06] But one of the things I wanted to do was actually get a feel for how the CPU temperatures were in that system over time, |
[1415.06 → 1416.80] because I've got a couple of hard drives in there. |
[1417.44 → 1421.16] And, you know, if things are getting toasty, hard drives tend to die pretty quickly. |
[1421.40 → 1429.50] So what I wanted to do was monitor that VMware venter instance with Prometheus and have it alert me if there was a problem. |
[1429.84 → 1430.06] Nice. |
[1430.06 → 1433.10] I ended up using this thing called VMware Exporter. |
[1434.16 → 1437.46] And there's a link to the Docker Compose snippet in the show notes. |
[1438.38 → 1447.00] And this thing, basically what it does is it logs into sphere's API every pre-prescribed interval. |
[1447.12 → 1449.26] I think I've got it set to like 15 seconds or something. |
[1450.04 → 1453.94] Scrapes a bunch of data available through the venter API, |
[1454.46 → 1457.44] which includes stuff at the host level like CPU temperature, |
[1457.44 → 1462.58] but also includes the PMI data like the voltages and all that kind of stuff. |
[1462.64 → 1466.08] So it's pretty granular and pretty detailed if you want to know that stuff. |
[1466.08 → 1469.70] And then it imports it into the Prometheus real-time database. |
[1469.70 → 1475.64] And from there, once it's in Prometheus, I can query it using Grafana or whatever else I want to. |
[1476.34 → 1481.90] And I use a container called Prior der, I think is how you say it. |
[1482.02 → 1483.52] Prior der VMware Exporter. |
[1484.22 → 1486.76] Super simple little Python app and just does the job. |
[1486.76 → 1489.74] What kind of hardware utilization, I guess, is what I'm asking. |
[1489.94 → 1493.28] So if I have a system, I have an existing Leon box here at the studio. |
[1493.90 → 1497.40] It's probably about 30% utilized CPU-wise and memory-wise. |
[1497.92 → 1499.36] Do I need something more powerful? |
[1499.48 → 1501.62] Do I have enough headroom for running these kinds of services? |
[1502.04 → 1504.16] Prometheus itself is pretty lightweight. |
[1504.90 → 1508.94] You know, with any real-time metrics gathering system like this, |
[1509.02 → 1512.34] we'll probably want to put it on some kind of flash-based storage media. |
[1512.34 → 1517.56] So, you know, you'll need a few gigabytes scratch disk for that to use. |
[1517.80 → 1521.76] But then once you've done that, you need to create what are called scrape jobs, |
[1522.02 → 1525.14] which is how you tell Prometheus what to look for |
[1525.14 → 1528.56] and what to kind of gather in terms of metrics on remote systems. |
[1529.08 → 1533.70] You configure that in my favourite language of them all in a YAML file. |
[1534.46 → 1538.42] And once you've done that with Prior der, you kind of set it, forget it, |
[1538.46 → 1540.36] and never worry about it ever again. |
[1540.36 → 1542.46] And then how long does it collect for? Forever? |
[1543.30 → 1544.44] How long do you do it for? |
[1545.02 → 1546.28] I haven't really worried about it. |
[1546.38 → 1546.98] I just sort of... |
[1546.98 → 1547.32] Forever! |
[1547.80 → 1549.16] Keep them forever, yeah. |
[1549.24 → 1551.88] I mean, I guess if you're in an enterprise situation |
[1551.88 → 1556.14] where you're scraping 10,000 servers, that might be an issue, but I'm not, so... |
[1556.14 → 1558.28] Yeah, so maybe not a big deal. |
[1558.40 → 1560.06] You could be. That'd be kind of fun, wouldn't it? |
[1560.10 → 1561.68] Go back 10 years and look at stuff. |
[1562.42 → 1562.78] Oh! |
[1563.38 → 1565.72] Well, I mean, look, this is how these big guys do it. |
[1565.72 → 1570.38] I mean, Kubernetes clusters the world over are monitored by Prometheus. |
[1570.94 → 1574.38] OpenShift actually ships a bunch of stuff with Prometheus built in. |
[1574.56 → 1578.28] So this is one of those technologies that if you're thinking about getting into DevOps |
[1578.28 → 1582.08] and that kind of space of, you know, site reliability engineering, |
[1582.80 → 1586.48] these are the sorts of tools that you should be mucking about within your home lab |
[1586.48 → 1588.94] and trying to figure out how they work and understand them |
[1588.94 → 1592.02] because you'll be able to take that knowledge and use it to get a better job. |
[1592.02 → 1593.36] Yep, great skill to have. |
[1593.90 → 1594.98] I just like the idea. |
[1595.04 → 1597.32] I just like the idea of picturing you like 10 years from now, |
[1597.40 → 1603.72] looking back at some graph analogs and just kind of getting nostalgic from graphs and logs. |
[1604.16 → 1607.98] Yeah, my CPU is 33 degrees on May the 5th, 2021. |
[1608.34 → 1609.06] Hey, great. |
[1609.22 → 1610.80] Ah, I remember that, yeah. |
[1611.16 → 1612.26] Those were the days. |
[1615.22 → 1616.30] Cloudfree.shop. |
[1616.30 → 1618.88] Head over there and buy from a community member |
[1618.88 → 1624.58] who has built a business out of creating devices that are ready to go without any cloud restriction. |
[1625.04 → 1626.92] And they posted an update in February. |
[1627.80 → 1629.90] And you got a mention in here. |
[1630.06 → 1630.72] He says... |
[1630.72 → 1631.08] Did we? |
[1631.38 → 1632.24] How cool is that? |
[1632.46 → 1632.92] He says, |
[1633.00 → 1635.68] I need to also thank Alex and the self-hosted podcast. |
[1635.92 → 1638.58] I've listened to Jupyter Broadcasting podcast for over seven years. |
[1638.92 → 1642.82] And without our partnership, Cloud free would not have grown nearly as fast as it did. |
[1643.12 → 1646.12] And I got to say, it hit me in the feels because the picture they have in that |
[1646.12 → 1649.68] looks a lot like what we're doing for the Jupyter Garage shipping right now. |
[1650.20 → 1654.84] And I just totally realized, like, I'm learning a lot of the same stuff that they had to learn |
[1654.84 → 1656.68] to do Cloud free right now. |
[1657.14 → 1660.38] And so I felt very like I had a kinship all of a sudden. |
[1660.54 → 1662.86] And I thought that was really great that they had a really... |
[1662.86 → 1663.58] This is a perfect... |
[1663.58 → 1664.68] This is a really positive update. |
[1664.76 → 1667.16] It looks like things are going really well for Cloud free. |
[1667.28 → 1668.68] And that's why we like to mention them. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.