| Anonymous | Login | Signup for a new account | 2013-05-21 03:55 EDT |
| Main | My View | View Issues | Change Log | Roadmap |
| Viewing Issue Simple Details [ Jump to Notes ] | [ View Advanced ] [ Issue History ] [ Print ] | ||||||
| ID | Category | Severity | Reproducibility | Date Submitted | Last Update | ||
| 0000332 | [Nagios Core] Checks | major | always | 2012-05-08 21:17 | 2012-05-14 05:40 | ||
| Reporter | eschoeller | View Status | public | ||||
| Assigned To | estanley | ||||||
| Priority | normal | Resolution | fixed | ||||
| Status | resolved | Product Version | |||||
| Summary | 0000332: Double quotes in check_command definition break functionality | ||||||
| Description |
We upgraded one of our nagios core environments today from v3.3.1 to 3.4.0 and started to experience large scale instability in many of our service checks. The symptoms varied between the various plugins but some example output: "Name or service not known" "No data was received from host!" "libexec/check_nrpe: invalid option -- w" Many of the plugins that functioned perfectly in a v3.3.1 environment were completely broken in v3.4.0 We eventually determined that we had been using double quotes in many of these check_command definitions, such as: check_https!"HTTP/1."!"443"!/ or similar. We found switching all of these double quotes to single quotes (or removing them entirely where unnecessary) solved all of the strange problems. If double quotes are now interpreted differently or not supported, a check of the configuration should indicate a problem ! |
||||||
| Additional Information | We are continuing to monitor the performance of this infrastructure, but since the upgrade (and config fix) we are seeing a very distinct change in the frequency of service checks being executed. | ||||||
| Tags | No tags attached. | ||||||
| Nagios Version | 3.4.0 | ||||||
| OS | RHEL | ||||||
| OS Version | 4 | ||||||
| Attached Files |
|
||||||
|
|
|||||||
Notes |
|
|
(0000423) whaity (reporter) 2012-05-09 10:24 |
I have the same issue. CentOS 6.2, Nagios 3.3.1 -> Nagios 3.4 upgrade. All work fine before. The only plug in I have that uses double quotes is the check_snmp_printer plug in. I have tested the command version and all works fine. If I run via Nagios I get a "Invalid check specified by -x parameter." I have tried single quotes but I get them same problem it seems. |
|
(0000424) prandal (reporter) 2012-05-09 12:31 |
It also breaks check_mk because of its use of definitions like this: # Calling check_mk with precompiled checks define command { command_name check-mk command_line python /var/lib/check_mk/precompiled/"$HOSTNAME$" } |
|
(0000425) tolim (reporter) 2012-05-09 12:59 |
It broke check_http and check_by_ssh here among other things..same symptoms. |
|
(0000426) jangs (reporter) 2012-05-09 23:36 |
Same error for check_by_ssh command /usr/local/nagios/libexec/check_by_ssh: invalid option -- 'w' Nagios 3.3.1 -> Nagios 3.4 upgrade OS: Debian OS Version: 6.0.4 |
|
(0000427) prandal (reporter) 2012-05-13 14:42 edited on: 2012-05-13 16:32 |
In checks.c, replacing if(!strpbrk(processed_command,"!$^&*()~[]|{};<>?`")){ with if(!strpbrk(processed_command,"!$^&*()~[]|{};<>?`\\'\"")){ fixes this issue. An appropriate char *temp_buffer declaration in the embedded perl compile path is also needed if checks.c is to compile with the embedded perl option turned on. Attached checks.diff fixes the issue. |
|
(0000428) estanley (administrator) 2012-05-14 05:40 |
Reverted patch that caused the problem (issue 0000086) |
| Mantis 1.1.7[^] Copyright © 2000 - 2008 Mantis Group |