Pingdom and FTP service checks... - UK Web Host Chat, Hosting Forum, Web Hosting Forum

LeaUK · 3rd July 2009

Haven't been around much as having a bit of a life crises...but hope you don't mind me asking for a little assistance?

We understand that Pingdom doesn't support a full FTP check (only a network port 21 open,) therefore if the FTP service fails no warning is given.

So I've been advised to use their 'HTTP Custom check' to call a script which accesses the FTP service and returns the result in xml, however this means that FTP uptime is now dependant on HTTP services (as far as I can see), which isn't really what I was hoping for.

But I note that their Network check allows 'string to send' and 'string to expect', so I'm wondering if this could be used somehow.

i.e. FTP servers generally respond with their name etc...

I suspect this would be good enough compared to actually logon to the FTP service with user credentials as it at least proves that the FTP service is running. We see our service fail once every 6-9 months for unknown reasons!

Any thoughts much appreciated.
Lea

JonWitte · 3rd July 2009

Custom connect to port 21 should give you what you need as a response - telnet {your ftp server} 21 will show the string to check for.

Pete-CCS · 4th July 2009

If you are checking for the header only take care - we got caught out around beginning of last year - the FTP server was responding with the header - but not accepting logons - We now test by logging on and getting a small file - Much more "proof" to my mind.

othelloRob · 4th July 2009

Quote:

Originally Posted by PeteK

We now test by logging on and getting a small file - Much more "proof" to my mind.

putting a file might be even better - we've had issues with FTP allowing gets & deletes but only puts files with a 0 byte length before ...

Pete-CCS · 5th July 2009

Quote:

Originally Posted by othellotech

putting a file might be even better - we've had issues with FTP allowing gets & deletes but only puts files with a 0 byte length before ...

Agreed Rob - probs should have stated we actually delete / put and get the same small file..

LeaUK · 7th July 2009

Hi Guys

Many thanks for the info and suggestions. I agree that a GET/PUT etc, would be far more revealing, however trouble is Pingdom only expect to receive a string, so I think I'm left with no choice than to check the header (thanks for the telnet trick btw). It's a significant improvement over just checking the ports.

Cheers all
Lea

LeaUK · 7th July 2009

There, all done. Magic string checking...all implemented!

Thanks guys, as I hate it when I've told all my clients how we monitor services at 1 minute intervals, only to find out that, well - we're not!

Lea

LeaUK · 16th July 2009

Just for info, I've recently added a raft of checks since discovering the 'return string' feature, even checking availability of our Control Panel running on https. OK, it's not full authentication as Pingdom do not support https login checks, but it at least informs us that the pages are rendering 'correctly'.

We had a couple of minor issues lately with links to SmarterStats displaying the Windows Authentication box rather than rendering the asp page login (turned out to be inconsistencies with Plesk domains and file ACLs of IIS websites) and hated the fact that our client's reported this to us before we knew - not being a regular to login to this service

So these new checks have given us significantly increased confidence that all is ok...

Oh and an opportunity to tell our clients how we have increased our monitoring services and how more responsive we can be now...nothing wrong with turning a minor inconvenience into something positive

Lea

Pete-CCS · 16th July 2009

Still think you should just buy Hostmon (or rent a bit from someone else ahem ahem) and then you can check all these without paying a small fortune to Pingdom...

It was quite possibly the best purchase we ever made - presently have 120k sensors on it. Very handy...

Tim - DR · 16th July 2009

Quote:

Originally Posted by PeteK

Still think you should just buy Hostmon (or rent a bit from someone else ahem ahem) and then you can check all these without paying a small fortune to Pingdom...

It was quite possibly the best purchase we ever made - presently have 120k sensors on it. Very handy...

How is hostmon any different/better to just running nagios Pete ?

You can do a lot more installing nagios on a box (or even a vps) than with pingdom and probably for less money ( just server/vps cost)

Pingdom is creating for monitoring the absolute basics but not for anything indepth.

Pete-CCS · 16th July 2009

Tim - Not saying it is - just suggesting certainly in our case - that Hostmon is very good. Ok, I don't mean it checks from 3983943 locations like Pingdom, but its off of our network and I know how it gets there and happy with what it checks - I just find it very friendly and usable. I also think the remote sensors are excellent - don't know if Nagios has any - don't recall it did last time I checked - and the ability to run jobs locally on a machine with a single encrytped port open give us the option (for example) to put it on a customers server (i.e. SBS or similar within a business) then spead check all their switches are there, scrape the logs, ping their WAP etc etc as well as checking their version of AV, Exchange queues, services etc etc...

LeaUK · 28th August 2009

Pingdom is fine for our few core servers and in my opinion is good value, however in ‘another life’ I use Server's Alive and maybe looking to change this. I monitor around 300 events across 60 servers but find the limitations of SA starting to show:

SA crashing causing report data loss

Maintenance of SA is horrible.

However it's very inexpensive

Slow check times - i.e. I'm nearly maxing out at a 5min polling cycle due to the number of events being monitored and that these are effected sequentially.

Wonder if HM carries out multithreaded checks in order to increase checking throughput?

But, the SA crashing issue is cause for serious concern now and I may want to move to an alternative.

Anyone used both? And what's your thoughts?

Thanks
Lea

Pete-CCS · 28th August 2009

Why not do it in house, or off-net on a dedi/vps?

domtaylor · 28th August 2009

We use Alertra and find it excellent - not especially cheap but it's never been slow/wrong.

LeaUK · 28th August 2009

Quote:

Why not do it in house, or off-net on a dedi/vps?

I do on my 60 or so servers, but for my other network (completely independent 20 sensors) a 3rd party like Pingdom is fine.

I'd be interested in people's experience of Host Monitor.

Lea