[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Variable / continuous ping / network test (intervals)
- To: <hobbit (at) hswn.dk>
- Subject: Variable / continuous ping / network test (intervals)
- From: "SebA" <spa (at) syntec.co.uk>
- Date: Thu, 13 Nov 2008 18:14:23 -0000
- Thread-index: Aci3Hn3yJSSLcI3lQwOHxIQz7ye7XyOlxZEg
I suppose a feature request for 4.3.0 (or 5.0) since I don't believe it
exists yet, is to have an flag in bb-hosts that tells bbnet how often to
test (or report tests for) each host. Some tests, e.g. on the LAN, are fine
being tested every 5 minutes, but it would be good to know when
interconnects go down or suffer packet loss (a) more quickly, and (b) for
shorter periods not detected by the 5 minute test unless we were lucky to
test at the right time. However, there's no need to test every single host
more frequently, which could slow down the time it takes for bbtest-net to
report.
Additionally (ideally): I would like to know results of continuous pings.
Rather than just sending a few pings every x minutes and seeing the
performance of those, it would be nice to have continuous, or at least pings
that are continuous for the duration specifed in the new bb-hosts flag I
have suggested above, e.g. at 5 or 10 second intervals (so as not to
overload routers or links), and to report the results of these tests every x
minutes (again, ideally, defined by the new tag I have suggested above).
OK, the result here is more than 5 minutes, but e.g. this Linux ping command
sends a ping request every 5 seconds, supresses the output and quits after
60 seconds or so:
$ ping -i 5 -q -w 60 xxx.xxx.xxx.xxx
PING xxx.xxx.xxx.xxx (xxx.xxx.xxx.xxx) 56(84) bytes of data.
--- xxx.xxx.xxx.xxx ping statistics ---
2360 packets transmitted, 2359 received, 0% packet loss, time 11795758ms
rtt min/avg/max/mdev = 1.350/2.078/61.857/3.262 ms
I imagine I can't use that ping command in the FPING environment variable
since there'd be concurrency issues. But if there was a way of doing
something similar, we could really see what the performance on our links is.
(It doesn't like there is the equivalent to the -w ping option in fping,
although that can be approximated to a reasonable extent with -c or -C
(count) given a known -i (interval).)
Kind regards,
SebA
_____
From: Josh Luthman [mailto:josh (at) imaginenetworksllc.com]
Sent: 16 May 2008 07:27
To: hobbit (at) hswn.dk
Subject: Re: [hobbit] ping intervals
Pretty sure that's
[bbnet]
ENVFILE /home/shire/server/etc/hobbitserver.cfg
NEEDS hobbitd
CMD bbtest-net --report --ping --checkresponse --concurrency=48
#CMD bbtest-net --report --ping --checkresponse --concurrency=64
#CMD bbtest-net --report --ping --checkresponse
LOGFILE $BBSERVERLOGS/bb-network.log
INTERVAL 5m
in etc/hobbitlaunch.cfg
On Fri, May 16, 2008 at 2:22 AM, Gavin Leonard <gleonard (at) progrexion.com>
wrote:
I have some servers that it is critical that I know they are down within at
the most a minute to two minutes. hobbit polls every 5min if I remember
right? How do I decrease the polling intervals so I can catch these
connection problems quicker? Thanks in advance.
Gavin Leonard
cid:image001.gif@01C856AD.922EF120
Director, Systems-Network Engineering
T
801-828-1735
F
801-828-1704
E
<mailto:gleonard (at) progrexion.com> gleonard (at) progrexion.com
Research | Marketing | Sales Generation
<http://www.progrexion.com/> www.progrexion.com
This email and its contents are confidential. If you are not the intended
recipient, delete this email and do not use or disclose the information
within this email or its attachments. Thank you.
--
Josh Luthman
Office: 937-552-2340
Direct: 937-552-2343
1100 Wayne St
Suite 1337
Troy, OH 45373
Those who don't understand UNIX are condemned to reinvent it, poorly.
--- Henry Spencer