New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Yeah i could see what you mean having something like this behind your firewall is nice but i think it also hurts if there is a planned revenuse stream with out asking for a lot of momney or problems with having to give away your source code or use things like ioncube. Something that i am sure that is out of scope but would be nice is to allow these guys to setup a vm instance for you add it to ur vpn and they deploy the code there this way everyone can be happy.
only thing not updating on the site for me is data. 0.00KB/s
Nice simple to use service. Easy to install and shows all the basic metrics I'm looking for.
My only complaint is that there is no public status page, no other issues with it than that.
Thanks! Everything seems to be running silky smooth the only thing missing is setting up the Monthly/Daily/Hourly thingy by default, but I've already mentioned that to you in email.
Is there any way to donate?
I would love support for Pushover!
Offering a self-hosted version comes with a lot of additional support work and is unfortunately not an option for now. I also think there are much more advanced solutions for experienced system administrators. We aim to be quick and simple to use which makes a web application the perfect fit. We will however tweak the alert system over time and let you set custom threshold for all three resource types.
As long as your network interface and graphs are working this shouldn't happen. It does however show it only for the last 3 minutes so maybe you just don't have any traffic on your server right now?
Something like that including an API will come eventually.
Thank you for the suggestion. I saw it on Twitter but didn't have time to reply yet. I will have a look at it.
I'm unable to open NodeQuery using my android tablet browsers... Got security warning (SSL cert) and it keep asking me for authorization. http://i.imgbox.com/O2gNc0qQ.png @Joe_NQ
One of my debian server is showing it is down , while it is not the case , it is up and everything is working as it should. should i reinstall the server or?. i did try to re-install it , but still the same
Very clean design, curious to see how it improves. Also wondering about the pricing to come along with a public page perhaps :-).
Only thing I really want is a public status page, or some sort of embed code.
Interesting, you were the first to report that issue. I made some changes to the SSL configuration. Is it working now?
Did the server work at all at some point or did you just receive down notifications? Maybe you can find something in the logs? Is the cron job running?
Public report pages including the API is on the list. I can't really say anything about the pricing yet. I hope it won't be more than a low end VPS for the smallest plan though.
@Joe_NQ Sounds good, liking what I see so far!
@Joe_NQ, nope.. still unaccessible and keep asking for authorization.
It could be related to SPDY which is enabled on the web application server. The site loads fine in Android 4.4 but fails on all lower versions. I will have a look at it again tomorrow. Thank you for your help
So every node is reporting data to your servers at the same interval to the millisecond basically?
Also, Is there any auto update feature for the agent?
The servers you add and install send metrics every 3 minutes. Of course it is not to the millisecond but the backend calculates the last interval to keep things in line.
Auto updating the agent would be a bad idea in terms of security and reliability. Imagine we release a faulty update, all your servers would stop working. With a manual installation you can update server by server and check if everything is in order.
This is something i was going to bring up also. Since your running on a cron jobs with */3 each server is going to start running this job pretty close to the top of the minute. Most servers will report back with in a normal amount of time 1-5 seconds. That is a lot of metrics to have come in all at once. If you have it going into a redis and then into the UI that is not to bad since redis can handle a lot but would not really recommend it. But also there is a chance the UI checks at the start of the new 3rd minute block and now marking a server down even though its not down and its still in the queue but near the end because of all the data that is comming int at once. You may wanna have the processes "sleep" for some random amount of time before or after it gets the metrics before sending it to you.
@wojons precisely
sleep $[ ( $RANDOM % 10 ) + 1 ]s
Excellent I prefer it that way. Especially considering I can review the source on GitHub
@Joe_NQ Do you record the IP posted along with the actual IP of the request? For example for security/privacy I may obscure my IPs via the agent.sh
Hello,
Just logged here to tell you that I've created an account there and am using it. I love it. I'd love it even more if it were to keep free, 10 servers, forever!
Now, you should get a way to create public pages, without any IP, for those who aren't programmers, such me, :-P
Keep working on it, I'll give you any feedback I think it's worth mentioning.
Do you have any webpage where we can post feature-requests?
I'd like to help more and suggest a few things that I have thought it would be worth to implement.
Edit: Thank you for the free service, though, I definitely love it.
@Joe_NQ
It seems to report incorrect Disk size of KVM machines.
Really cool service, love how simple it is.
@GaNi: disk size for KVM is reported correctly for me
Absolutely love it - thanks Joe.
The morris.js historical charts are especially beautiful and useful. The client-side code looks simple and logical. I like that you're sending the data over HTTPS.
Couple of small suggestions... maybe already on your plan:
cache the logging info client-side. Then the info is safe against any connectivity issues (or you having to reboot your servers). Also you could detect the fact that your server's then receiving a multiple-sample packet and somehow indicate connectivity issues on the graphs - I guess the logical thing might be to grey-out a portion of the graph line but I'm fairly sure morris doesn't support that. I notice you've got an availability measurement, so maybe you're doing something for this already (but I'm in the fortunate position of having 100% availability right now, so don't know)
a lot of the info doesn't change from packet to packet. Maybe only send CPU, OS, kernel etc if it differs from the last report. Although downsides are obviously more complex client code and having to deal with cases where a change might have been missed server-side - probably more trouble than it's worth.
the Network Usage graph is showing "MB since last sample" isn't it? I wonder if that's the most logical unit, esp since it's across a period of 3 minutes. Maybe go with Mb/s
I know in the PHP world the sight of anyone using base64 encoding scares a lot of people because of the frequency with which it's used for abusing hacked servers. JSON might be a clearer alternative? Also gives you a little more flexibility if you alter the info you're transmitting (I guess right now you'd have to check the version field in the packet & say "ok, it's version 0.7.3 therefore I know the packet contains an extra field and I know it contains xyz")
given that the installation script is making some modifications to the user's server, how about an uninstall option as well that'll remove the crontab entry, delete the directory, delete the user & group etc
for monetization, I'd be tempted to go with more frequent status updates (although you'd have to enforce that server-side given that the data is pushed from the client - otherwise anybody could change that themselves) and SMS alerts.
Beautiful job!
I agree, right now it is not a problem at all but might become one with more servers being monitored. The reason for occasional false down notifications is simply the threshold and delay between servers and the web service. I've modified the values and hope it will help the few users who were affected.
No, we do not record requests made from your servers. Not even in the access logs as it would be just too many. Only the data specified on Github is being stored.
We have a Uservoice page linked in the footer where you can post suggestions.
Before launching the beta I tested the script on so many KVM systems and never had a problem. Is it reporting too much disk space? Maybe there is misconfiguration on your node?
Thank you very much, that's quite the detailed feedback.
For now it is supposed to be as simple as possible. If your server is down or has a connectivity issues, no data is being send and the availability score is getting lower. There might be potential for improvement in the future.
You have a point but it would require additional requests and would complicate many of the processes in place. The data packets are relatively small so even in a network with higher bandwidth costs it shouldn't really be a problem.
Correct, the network usage graph is showing the data transferred in the last segment. When hovering over an interval point you can see the average in */s. In my personal opinion the actual amount of transferred data is more interesting especially when looking at the monthly graph.
We used JSON in an early version, but escaping all the different values and creating the markup manually got too messy. In the end I decided to just go with a base64 string. I think people like you who look at the code know that nothing bad is happening in the code. Anyone else probably won't care how the monitored data is being transferred as long as it works and is secured via SSL.
A simple uninstallation command is displayed as soon as you remove the server from the interface. I will create a help page soon that gives the user more information on that.
SMS alerts will definitely be available in the pro plans.
Thanks to everyone again for the awesome feedback
@Joe_NQ
I installed the script as instructed, did no changes. Here's what it looks like(on both of my KVMs)
But df -h shows up accurate.
@Joe_NQ excellent work so far - simple yet elegant. I would highly recommend considering an API.NQ sub domain/vhost as your going to want to segment the traffic and it will ease scalability down the road.
Yeah, @Joe_NQ what did you use to make it?
Looks like it is showing the SWAP instead of the disk? Could you post or send me the output of 'df -h'?
An API will definitely come, I just don't have an ETA yet.
The front end is completely done with Espresso for Mac, right now my favorite HTML editor. I wanted to switch to Sublime Text but didn't have the time yet to get into it. Before writing the markup I did a lot of concepts in Sketch. Does that answer your question?