this looks ace, so a serious question: is it widely used?
this is the first I've seen of it (seems true of other people commenting too), but it looks like there's some strong history (several acquisitions, then joined Linux Foundation in 2019)... so why hasn't it achieved greater visibility?
those questions are not asked to rain on someone's parade, but because going all-in on automation and workflow is a gamble as to whether that thing that glue is going to stick around (Linux Foundation suggests yes), and also is a skill that people want to learn and use (being widely adopted and something people can apply to other companies and roles would make this so, hence the questions)
Soooo I think I'm far enough removed from my old job to comment-
I worked at Adobe as a lead on a platform engineering team. Another team came to us pushing for us to adopt StackStorm for autoremediation of our infrastructure. We looked at it, and didn't see it as a good fit for us. The example we were given was "if httpd goes down, restart it". But:
- Systemd's built in restarts worked orders of magnitude faster than StackStorm
- At higher levels, the Kubelet, Kubernetes workload controllers, and cloud-native services provided fairly robust autoremediation capabilities
Despite this, we were still pressured to use it. I think some other team saw my team's success (we eventually ran the infrastructure for a majority of Adobe's services, and I believe that's still true) and wanted to hitch a ride on the rocket. But we never did.
Note the wording on that ADOPTERS.md line - that line only applies to one team at Adobe, a company with tens of thousands of employees and services across hundreds of products. I can tell you while I worked there, StackStorm did not seem to be widely used.
The website lists connecting various third party services together. I think that makes sense for quickly snapping together vendor services without having to figure out how to wire up their APIs. For cases where we were fully responsible for all the software involved, it was like trying to drive a car from the back seat with pieces of string.
> this looks ace, so a serious question: is it widely used?
I know some french cloud provider that uses it, and… I'm not sure they're happy with it. Issues with YAQL, a lot of weird bugs. They mentioned Temporal looking much cleaner.
it's been around a long time with serious usage, hn frontpage just doesn't typically provide a thorough market analysis perspective and indeed has many non-expert (in terms of gained familiarity with a correctly broad view of our field) members. still doesn't mean it's necessarily a great choice though
Love StackStorm ! I had never heard of it before my employer adopted it, despite having looked for exactly what it provides before.
Writing integrations is very easy, which makes it a great “hub” where events from different sources can be collected and watched, while miscellaneous tooling can be gathered behind the same UI and auth. My team is adopting it big time.
It’s not without warts and there are alternatives better suited for other environments, but I still believe StackStorm deserves more attention than it has received !
It is one of the few workflow managers that support cyclic graphs (e.g. ecflow, cylc, prefect, covalent more or less.
), instead of working only with DAGs like Airflow.
It is a feature common for weather and climate workflows, and I think other HPC and machine learning workflows too.
FYI if you need a cycle in Airflow you can stitch DAGs together with datasets. The simplest would be to just have a DAG update the same dataset that it consumes. Whether it actually corresponds with a dataset out there is immaterial, it can just be an event trigger.
st2 reminds me a lot of what we're building at https://github.com/keephq/keep (i maintain this oss), there are a few concepts that i really love but the video in the repo really gives me a 90s feelin'
i can't really seem to understand, there's an "enterprise" (with ldap, etc.) version but couldn't find any pricing info
edit: well it's not focused on alerting but alerting is somewhat kinda "event driven"
I’ve used ST2 on and off for many years, and it blows the socks off folks at every place I introduce it. Unfortunately, some places have a strong resume-driven-development policy, so if it isn’t on the hip new cargo cult list of the day, nobody is going to have a look at it.
I was recently called a “boomer” for bringing it up. I really need a new job.
I don’t see a pricing page - this clearly looks like a product supported by a company. What’s the plan going forward ? Will it still be aviable in a few years ?
We got confused when researching StackStorm so to save you some digging : the software has been donated to the Linux foundation and is no longer supported by a commercial offering. See:
Development is currently quite active. Everybody’s risk analysis is different, but my assessment is that :
- open source means that we can contribute the fixes and improvements we need
- not being dependant on a commercial product protects us from events outside our control : what if the company goes under and does not release their code ? What about licensing / pricing changes ?
- it follows that in the worse case scenario where the project is completely abandoned, we have the code base and opportunity of maintaining it
- others are in the same position, which strengthens the incentive to collaborate on maintenance
That is not to say I’d systematically reject commercial products in favour of FOSS, just that both have different risk profiles that may or may not fit a given situation :)
There isn't much advantage over github actions. It is from the ansible and puppet era. You also have to be well verse with ansible or puppet to operate properly with stackstorm.
If you are starting out with event driven devops automation, you would choose github actions.
this is the first I've seen of it (seems true of other people commenting too), but it looks like there's some strong history (several acquisitions, then joined Linux Foundation in 2019)... so why hasn't it achieved greater visibility?
those questions are not asked to rain on someone's parade, but because going all-in on automation and workflow is a gamble as to whether that thing that glue is going to stick around (Linux Foundation suggests yes), and also is a skill that people want to learn and use (being widely adopted and something people can apply to other companies and roles would make this so, hence the questions)