Xymon Mailing List Archive search

weird http test issue on 4.3.21

4 messages in this thread

list Bruce Ferrell · Wed, 4 Nov 2015 23:00:08 -0800 ·
I have a set of name based virtual hosts that I am testing.  the hosts.cfg for this looks like this:

page Web_Sites "Web Sites"
group-only http|info|trends
WWW.XXX.YYY.ZZZ  site1.org    # noconn  http://site1.org/sitemap.xml
WWW.XXX.YYY.ZZZ  site2.com    # noconn http://site2.com/sitemap.xml
WWW.XXX.YYY.ZZZ  site3.com    # noconn  http://bbjsails.com/sitemap.xml
WWW.XXX.YYY.ZZZ  site4.com    # noconn  http://dhluxuries.com/sitemap.xml
WWW.XXX.YYY.ZZZ  site5.com    # noconn  http://ferrellsystemsengineering.com/sitemap.xml
WWW.XXX.YYY.ZZZ  site6.com    # noconn  http://plainjaneofficeworks.com/sitemap.xml
WWW.XXX.YYY.ZZZ  site7.org     # noconn  http://sg-creations.org/sitemap.xml

all of the sites but the first work.  I've tried, just for fun, re-arranging the order of the sites... No change.

I notice the XYMONNETSVCS set is : smtp telnet ftp pop pop3 pop-3 ssh imap ssh1 ssh2 imap2 imap3 imap4 pop2 pop-2 nntp ftps telnets smtps submission msa pop3s imaps nntps ldap
ldaps rsync bbd clamd spamd svn oratns qmtp qmqp amqp amqps vnc cupsd ajp13 rdp netbios-ssn snpp ircd lpd
 
There doesn't seem to be any http test defined.

The apache logs for the failing site show no access to sitemap.xml

Anyone have any thoughts?
list Japheth Cleaver · Thu, 5 Nov 2015 05:44:03 -0800 ·
quoted from Bruce Ferrell

On Wed, November 4, 2015 11:00 pm, Bruce Ferrell wrote:
I have a set of name based virtual hosts that I am testing.  the hosts.cfg
for this looks like this:

page Web_Sites "Web Sites"
group-only http|info|trends
WWW.XXX.YYY.ZZZ  site1.org    # noconn  http://site1.org/sitemap.xml
WWW.XXX.YYY.ZZZ  site2.com    # noconn http://site2.com/sitemap.xml
*skip*
quoted from Bruce Ferrell
all of the sites but the first work.  I've tried, just for fun,
re-arranging the order of the sites... No change.

I notice the XYMONNETSVCS set is : smtp telnet ftp pop pop3 pop-3 ssh imap
ssh1 ssh2 imap2 imap3 imap4 pop2 pop-2 nntp ftps telnets smtps submission
msa pop3s imaps nntps ldap
ldaps rsync bbd clamd spamd svn oratns qmtp qmqp amqp amqps vnc cupsd
ajp13 rdp netbios-ssn snpp ircd lpd

There doesn't seem to be any http test defined.

The apache logs for the failing site show no access to sitemap.xml

Anyone have any thoughts?

Hi,

'http' as a test shouldn't need to be listed in XYMONNETSVCS explicitly
since it's handled by its own conversation mechanism, and a problem there
would end up affecting all 'http' tests, not just that one.

If it's following the hosts.cfg entry around, it's probably xymonnet
really seeing it as in bad state.

- What's the error message given in the http test for the one that's failing?

- Can you get to the site using wget from your monitor server? Ruled out
network/DNS?

- Are you using SNI on this server, or is there anything distinct about
that one vhost? Try toggling --sni/--no-sni

- Try running xymonnet in debug mode (just that one host) and see what it
gives you:

xymoncmd xymonnet --debug  --ping --checkresponse --test-untagged site.org


HTH,

-jc
list Bruce Ferrell · Thu, 5 Nov 2015 07:41:40 -0800 ·
quoted from Japheth Cleaver
On 11/5/15 5:44 AM, J.C. Cleaver wrote:
On Wed, November 4, 2015 11:00 pm, Bruce Ferrell wrote:
I have a set of name based virtual hosts that I am testing.  the hosts.cfg
for this looks like this:

page Web_Sites "Web Sites"
group-only http|info|trends
WWW.XXX.YYY.ZZZ  site1.org    # noconn  http://site1.org/sitemap.xml
WWW.XXX.YYY.ZZZ  site2.com    # noconn http://site2.com/sitemap.xml
*skip*
all of the sites but the first work.  I've tried, just for fun,
re-arranging the order of the sites... No change.

I notice the XYMONNETSVCS set is : smtp telnet ftp pop pop3 pop-3 ssh imap
ssh1 ssh2 imap2 imap3 imap4 pop2 pop-2 nntp ftps telnets smtps submission
msa pop3s imaps nntps ldap
ldaps rsync bbd clamd spamd svn oratns qmtp qmqp amqp amqps vnc cupsd
ajp13 rdp netbios-ssn snpp ircd lpd

There doesn't seem to be any http test defined.

The apache logs for the failing site show no access to sitemap.xml

Anyone have any thoughts?
Hi,

'http' as a test shouldn't need to be listed in XYMONNETSVCS explicitly
since it's handled by its own conversation mechanism, and a problem there
would end up affecting all 'http' tests, not just that one.

If it's following the hosts.cfg entry around, it's probably xymonnet
really seeing it as in bad state.

- What's the error message given in the http test for the one that's failing?

- Can you get to the site using wget from your monitor server? Ruled out
network/DNS?

- Are you using SNI on this server, or is there anything distinct about
that one vhost? Try toggling --sni/--no-sni

- Try running xymonnet in debug mode (just that one host) and see what it
gives you:

xymoncmd xymonnet --debug  --ping --checkresponse --test-untagged site.org


HTH,

-jc
Thanks for the tip!  I was trying to run the debug mode and failing 
miserably.

what I found was that I had an earlier entry for site1.org... without 
the http test.  That entry took precedence over the "Web Sites" entry... 
Oddly, there are actually other tests specified under web sites and they 
executed.  It's like that entry was partially de-duplicated and http 
skipped.
list Japheth Cleaver · Thu, 5 Nov 2015 13:40:04 -0800 ·
quoted from Bruce Ferrell
On Thu, November 5, 2015 7:41 am, Bruce Ferrell wrote:
Thanks for the tip!  I was trying to run the debug mode and failing
miserably.

what I found was that I had an earlier entry for site1.org... without
the http test.  That entry took precedence over the "Web Sites" entry...
Oddly, there are actually other tests specified under web sites and they
executed.  It's like that entry was partially de-duplicated and http
skipped.
Hmm. From a xymonnet perspective, I believe it should actually combine all
of those together into a unified test protocol (while xymongen does have
the possibility of getting confused about which definition is canonical).

I haven't run into that issue before, but it's possible there's a latent
bug that needs to get investigated here...


Thanks for the pointer.

-jc