How to monitor how many seconds server is down?
- Thread starter ICPH
- Start date
ICPH
Member
i dont wish to copy any files on that server that is expected to be down, so i think this will not work
This might be a good option: http://oss.oetiker.ch/smokeping/
Never used it, but I've heard good things. The only problem is that is based on a single probe so if the ping gets dropped it might not be because your VPS is offline but maybe the traffic between your VPS and the SmokePing server is experiencing packet loss. You can probably grab 2 or 3 cheap yearly VPSs and run SmokePing from all of them to have a more accurate report.
Never used it, but I've heard good things. The only problem is that is based on a single probe so if the ping gets dropped it might not be because your VPS is offline but maybe the traffic between your VPS and the SmokePing server is experiencing packet loss. You can probably grab 2 or 3 cheap yearly VPSs and run SmokePing from all of them to have a more accurate report.
GIANT_CRAB
New Member
I don't recall pingdom being able to do "seconds" in their probe. They do "minute" probing
I believe if you use the premium edition of Pingdom, this notifies you of any instability.
I don't know any reasonably priced service that would allow you to monitor every 1 second. If 1 minute is enough go for Pingdom, Hyperspin or any other. You can always write own script that will do it every 1s.
You can do second based monitoring with http://www.redotheweb.com/uptime/
ICPH
Member
During installation i got error "-bash: npm: command not found"
yum install npm wants to install around 90 packages of 17M which is quite alot.
Last edited by a moderator:
joesagona
New Member
Agreed, Zabbix will give you the most out of an open source tool for Linux. You can also try Nagios, but it is criticized for being very difficult to configure and create a lot of unnecessary noise. If you are looking for even more detail in downtime/ uptime and VPS performance overtime, BigPanda is a fairly new tool that integrates with Zabbix and will give you all the details, performance metrics, analytics, etc. you would want to look for.
I understood the OP expected a somewhat long downtime caused by a specific one time event and notified in advance (e.g. physical server rack/data center migration). It sounds an use case where precision is not a requirement and any dirt cheap solution is ok.
BTW you can't say a "server is down" monitoring it from a bunch of remote servers. A server that doesn't reply remote requests from a given route means exactly that.
Last edited by a moderator: