Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Show HN: SeaLion 2 – Linux Server Monitoring, Alerting and Debugging Tool (sealion.com)
32 points by treskot on Sept 18, 2014 | hide | past | favorite | 43 comments


I've got two problems here.

1) Cloud based. No, no, no, absolutely not, no way, no how. I am not hooking up server farms to the internet. Monitoring systems stay behind the firewall. Please come up with a self-hosted version.

Moreso when:

2) They log directly into the system. Unprivileged user or not, if you've got shell on my box, "you" being a random company on the internet, it's not my box anymore. Someone hacks you and by extension they've hacked me.


I agree with you, though other people may be entirely trusting of the SaaS in question or not care that much about security.

Commando.io is basicly the same service command-execution wise, however, it also offers the possibility for customers to run commando.io self-hosted (presumably enterprise pricing). Having a SaaS I presume, which is working (and successful) can be a boon to future enterprise sales that are concerned with the risks you mentioned.

I wish that all SaaS dealing with access to remote servers have some kind of (more up-front) disclaimer, noting that no matter how secure the service is advertised it will not be responsible for future mishaps/breaks/leaks.


It wasn't clear from reading the homepage, but I was hoping this only sent data OUT via HTTP and didn't require giving access to the system at all. If they actually require access to my system, that's a little disappointing.


I'd like to second this. Add a self hosted version and you have a new customer.


Immediately interested by the title of this post as I'm using a competing product (Scout App) right now, clicked through and subsequently left the site within 10 seconds.

That is a terrible website considering your target market are UNIX geeks.


What's wrong with Linux geeks looking at well designed websites?


I'm a UNIX geek and I too love "well designed websites" - and imho that site isn't a well designed website.

The majority of us are probably using a touchpad on our laptops and the scrolling thing is just horrible in that case.

The target customer will expect easily laid out information (as this is a web-based monitoring service) and if they see this "at the front door" then they'll run a mile.


I'm also a UNIX geek, and I would have to disagree with your opinion.

The design looks solid. I don't know anyone that has troubles scrolling on their laptop these days. I found that quite bizarre to hear. With daily use, you can be nearly as nimble as using a real mouse.

Could you give some suggestions on what you would change? I'm curious how you would display that much data on the screen in a clean manner.

Not trying to argue here, genuinely interested in improving my UI/UX knowledge. Quite useful when building webapps these days.


I agree. I opened it on my mobile device and the page was immediately moving and making all kinds of transitions. There's not really any instructions, nor are there any visible controls. It barely even registers as a functional website, so I don't even get to read about the product.


I've been a pretty big fan of Scout (ScoutApp) for this. To me, Scout's killer feature is the ease with which you can write a custom plugin. And, if that's too much trouble, just use the Generic JSON URI plugin and have your code/cron/whatever dump a json file with whatever you want to track.

This is quite a bit cheaper though.


tl;dr this tool is a cron for your monitoring scripts that reports to a cloud service which then gives you dashboards

Product information that I wish was on one page but instead is spread across five:

  - by default, agent only collects these [linux] stats
    - 1/5/15 min load average
    - cpu usage per cpu
    - memory usage (total,res,virt,cached)
    - network reads per second (RX?)
    - network writes per second (TX?)
    - disk reads per second (only #, so probably iops)
    - disk writes per second (only #, so probably iops)
  - collects output from monitor scripts [that you have to write]
  - sends output to sealion cloud via ssl
  - 'Your password is encrypted' (???)
  - have to edit a couple of files to make it collect info as root
  - have to create your own alerts
  - all data is erased after 3 (free) 15 (paid) or 45 (paid) days
  - features:
    - Dashboard & Charts
    - Alerts
    - Daily Digest
    - Raw output [from your monitor scripts]
    - Quick Setup
    - Enterprise Scale (???)
    - Time Machine (Server data is recorded for a week)
      (this is diff than 3/15/45 days reported above???)
    - Side-By-Side Server Comparison
    - Teams [access controls]
  - Pricing
    - FREE    2 servers  3 days data retention
    - $29/m   5 servers  15 days data retention
    - $49/m   10 servers 15 days data retention 
    - $249/m  50 servers 15 days data retention
    - $499/m  unlimited  45 days data retention
  - agent runs on your servers as unprivileged user
    - made with python 2.6
    - https://github.com/webyog/sealion-agent


I come from using New Relic for our server monitoring, and there are a couple things that I can see that I already love about Sea Lion, their realtime stats, and ad-hoc commands really opens up to allow any kind of monitoring. I would like to see the ability to rename servers instead of using their hostname.. as well as multiple stats on the page at once.. Other than that, this is really awesome!

Considering going on one of their paid tiers


Is there an equivalent of New Relics application performance monitoring? For example the view of time spent in script, DB, memcache, external, etc all in a stacked line graph?


You mean something like this?

http://www.appneta.com/images/graphics/slides/traceview/trac...

:)

Our languages support is on-par with NewRelic. We just recently release NodeJS instrumentation.

Disclaimer: I work for AppNeta. Our TraceView product does what you want and more (across servers, SOA).


Looks nice, but a bit pricey.


Yes, because our offering does more than what NewRelic does.


Well, I doubt it's because that, but because supposedly you have some large clients so you're not so desperate to become a mass-market tool. I personally find New Relic to be extremely expensive as well. The hardware cost of our servers is much less than our New Relic subscription. Also, there are some cheaper than New Relic or similar in cost solutions that do more than what New Relic does (AppDynamics, let's say).


Noone has really been able to knock new relic for the ubiquity, cross-language support, etc..

New Relic's server monitoring is a toy, however - that's why it's free.


Sorry to be the negative-nancy, but did Nagios (specifically XI) suddenly become hard or something?

Maybe I'm just an old neckbeard by now, but what's with this trend with cloud-hosted middle-man applications (read: expensive web 2.0 frontends) for standard software? If you can't install and configure your own server monitoring, what are you doing hosting your own servers? Am I missing something obvious here?


Though on one hand I'm inclined to agree with you regarding Nagios, on the other hand I disagree. What you're missing is the concept of developer empowerment. I am a sysadmin / operations engineer, and one trend I've noticed over the course of the past ten years is one toward developers venturing further and further into what could be labeled "traditional" systems administration. Do not underestimate developer demand. It's fueled many technological "movements" in the past ten years, infrastructure automation being a huge one.


I'm not going to get into the pros and cons of "should you hire a sysadmin", but I think the priorities of a lot of these businesses are really screwed up.

Fixed costs like an employee are scary even though services cost more and security isn't even a real concern.


I can't argue with you there. I'm fortunate that my current organization isn't this way, but I have worked at places where this has DEFINITELY been the case. It was mind boggling then and it's mind boggling now. The only explanation I can think of is the (at least perceived) speed and impatience of the modern customer / market. Maybe businesses feel a pressure to make these sacrifices? I'd even go as far as to say the marketing teams for these various "services" capitalize on this feeling of urgency, and try to perpetuate it culturally (e.g. "If you don't use <insert ultra insecure SaaS app here> your company is going to fall behind. Here, look at this list of all of the other companies that aren't yours that use our service!"


What if you're using AWS already? I couldn't find any word on CloudWatch integration.


Just sent you an email, but for anyone else who's wondering the same thing: I'll plug Scalyr, which is a server monitoring service with an option to import CloudWatch metrics. https://www.scalyr.com/solutions/import-cloudwatch


A shameless plug, I guess, but although I've seen Scalyr before, I will take another look at it today. Thanks!


Love the simplicity and the extensibility of running ad-hoc commands. Pricing looks okay, especially the unlimited tier as I hate per server pricing but $500 cap seems reasonable.


Spelling mistake in the very first screenshot in the very first visual when I clicked your link. Seperate -> Separate.


Good catch. Fixed it. Thanks!


Was very easy to install. But could not figure out how to monitor more than one indicator at the same time.


How does this work? Do I need to upload some a file on my server or do any special configuration?


Run the curl command once you sign up. Or, download the tarball and install it manually.


Generally speaking,

  rm -rf /
is a good starting point for those open to running random scripts from the Internet :)


Can you explain little bit more on "Debugging Tool"? How to use as Debugging tool?


Once you have custom commands set up, you can rely on alerts to zero-down on the issue and debug it with ease.

You can make use of historical data to go back in time, analyse and debug issues.


The scroll thing is very annoying. Can't you just have a list of product features?


More and more companies copy this blindly. There are cases when this works well, but here, it's more than just annoying.


An unfortunate case of designing for trends instead of designing for users.


This style is Powerpoint for the web.


The thing that gets me is the color contrast between "slides". Not very pleasing to the eye, unsettling really..


This can't be stated enough: everyone hates the scroll thing. Everyone. Even if people don't hate it, it sure isn't going to bring them more business. More the opposite.


Ah! Here's the list - https://sealion.com/features


i agree that the scrolling is very annoying. If it is on mobile even worst.


I didn't find it terrible until I got to the end I kept scrolling and ended up back at the beginning. That was a pretty terrible experience.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: