Xymon Mailing List Archive search

SSL test flapping after latest snapshot

4 messages in this thread

list Charles Jones · Thu, 20 Jul 2006 09:46:12 -0700 ·
I upgraded a rather old hobbit install that I have (I don't remember what version, it was an early beta), and afterwards an HTTPS test I had against a host started flapping (see attached screenshot).  Is there a recommended way for me to diagnose this problem?  Otherwise my next plan is to completely wipe the old install except for the RRD files, then build and install the latest snapshot again.

-Charles
list Charles Jones · Thu, 20 Jul 2006 11:51:55 -0700 ·
Okay so I went through with my plan to wipe everything and reinstall.  Now I have a new problem. The statuses for the hobbit server itself, specifically cpu, disk, files, memory. msgs, and procs have all turned purple.

I'm seeing these errors in my hobbid.log:
2006-07-20 11:55:51 Bogus status message contains no data: Sent from hobbitd
2006-07-20 11:56:52 Bogus status message contains no data: Sent from hobbitd
2006-07-20 11:57:54 Bogus status message contains no data: Sent from hobbitd

-Charles
quoted from Charles Jones

Charles Jones wrote:
I upgraded a rather old hobbit install that I have (I don't remember what version, it was an early beta), and afterwards an HTTPS test I had against a host started flapping (see attached screenshot).  Is there a recommended way for me to diagnose this problem?  Otherwise my next plan is to completely wipe the old install except for the RRD files, then build and install the latest snapshot again.

-Charles

list Henrik Størner · Thu, 20 Jul 2006 23:16:38 +0200 ·
quoted from Charles Jones
On Thu, Jul 20, 2006 at 09:46:12AM -0700, Charles Jones wrote:
I upgraded a rather old hobbit install that I have (I don't remember 
what version, it was an early beta), and afterwards an HTTPS test I had 
against a host started flapping (see attached screenshot).  Is there a 
recommended way for me to diagnose this problem?
What kind of error does it report for the failed tests ?

If you login as the hobbit user and run

   bbcmd bbtest-net --debug HOSTNAME

where HOSTNAME is the name of the flapping host, then it will dump
a lot of info about how it connects to the host and what happens during
the test. Comparing these for a failing and a succesful test could give
a clue about what happens.

The odd thing is that it's flapping like that. I could understand if it
would completely fail; I've seen that happen with a few SSL
implementations I test where upgrading the OpenSSL library caused it.


Regards,
Henrik
list Henrik Størner · Thu, 20 Jul 2006 23:30:50 +0200 ·
quoted from Charles Jones
On Thu, Jul 20, 2006 at 11:51:55AM -0700, Charles Jones wrote:
Okay so I went through with my plan to wipe everything and reinstall.  Now I have a new problem. The statuses for the hobbit server itself, specifically cpu, disk, files, memory. msgs, and procs have all turned purple.

I'm seeing these errors in my hobbid.log:
2006-07-20 11:55:51 Bogus status message contains no data: Sent from hobbitd
I *think* these are caused by a test going purple, but it is odd that
these messages have no data.

Could you try applying this patch ? It won't change the behaviour, but
it will log the host- and test-name of the status that it complains
about.


Regards,
Henrik

-------------- next part --------------
--- hobbitd/hobbitd.c	2006/07/20 16:06:41	1.250
+++ hobbitd/hobbitd.c	2006/07/20 21:28:41
@@ -957,14 +957,15 @@
 	dbgprintf("->handle_status\n");
  	if (msg == NULL) {
-		errprintf("handle_status got a NULL message for %s.%s, sender %s\n", -			  textornull(hostname), textornull(testname), textornull(sender));
+		errprintf("handle_status got a NULL message for %s.%s, sender %s, color %s\n", +			  textornull(hostname), textornull(testname), textornull(sender), colorname(newcolor));
 		return;
 	}
  	msglen = strlen(msg);
 	if (msglen == 0) {
-		errprintf("Bogus status message contains no data: Sent from %s\n", sender);
+		errprintf("Bogus status message for %s.%s contains no data: Sent from %s\n", +			  textornull(hostname), textornull(testname), textornull(sender));
 		return;
 	}
 	if (msg_data(msg) == (char *)msg) {
@@ -2395,7 +2396,7 @@
 					/* Count individual status-messages also */
 					update_statistics(currmsg);
 -					if (h && t && log && (color != -1) && (color != COL_PURPLE)) {
+					if (h && t && log && (color != -1)) {
 						handle_status(currmsg, sender, h->hostname, t->name, grouplist, log, color, downcause);
 					}
 				}