Troubleshooting » History » Version 11
Luke Murphey, 08/28/2019 05:48 PM
1 | 1 | Luke Murphey | h1. Troubleshooting |
---|---|---|---|
2 | 1 | Luke Murphey | |
3 | 1 | Luke Murphey | Below are some pointers to help in debugging the application: |
4 | 1 | Luke Murphey | |
5 | 1 | Luke Murphey | h2. Search for modular input errors |
6 | 1 | Luke Murphey | |
7 | 1 | Luke Murphey | Run a search for the following and see if the logs indicate a problem with the modular input: |
8 | 1 | Luke Murphey | |
9 | 1 | Luke Murphey | <pre> |
10 | 9 | Luke Murphey | index=_internal (sourcetype=web_availability_modular_input ERROR) OR (traceback AND NOT sourcetype=splunkd_ui_access) |
11 | 4 | Luke Murphey | </pre> |
12 | 4 | Luke Murphey | |
13 | 1 | Luke Murphey | h2. Search for REST handler errors |
14 | 1 | Luke Murphey | |
15 | 1 | Luke Murphey | Run a search for the following and see if the logs indicate a problem with the REST handler: |
16 | 1 | Luke Murphey | |
17 | 1 | Luke Murphey | <pre> |
18 | 1 | Luke Murphey | index=_internal source=*website_monitoring_rest_handler.log |
19 | 1 | Luke Murphey | </pre> |
20 | 2 | Luke Murphey | |
21 | 11 | Luke Murphey | h2. Input fails and generates a "ConnectionError" |
22 | 2 | Luke Murphey | |
23 | 2 | Luke Murphey | This can happen when the following is |
24 | 2 | Luke Murphey | |
25 | 2 | Luke Murphey | # The IP address is invalid (if referring to the host by IP) |
26 | 2 | Luke Murphey | # The domain name cannot be resolved |
27 | 2 | Luke Murphey | # The port number provided is invalid (if a port is specified) |
28 | 1 | Luke Murphey | # Network connectivity is lost or a firewall blocks the connection |
29 | 11 | Luke Murphey | |
30 | 11 | Luke Murphey | I have seen numerous of these issues which turned out to be due to environmental issues in the customer's network. I suggest starting with the assumption that the app is correct and that the network issues are real. |
31 | 11 | Luke Murphey | |
32 | 11 | Luke Murphey | h2. I'm seeing slow responses |
33 | 11 | Luke Murphey | |
34 | 11 | Luke Murphey | This is similar to the issue described above. I suggest examining the network connection for possible issues. Additionally, make sure that the host that is running the inputs is not overloaded (high CPU usage or high memory usage). |
35 | 3 | Luke Murphey | |
36 | 3 | Luke Murphey | h2. Troubleshooting modular inputs that don't appear in the Splunk Manager |
37 | 3 | Luke Murphey | |
38 | 3 | Luke Murphey | Look in the messages link in the top right of the page to see if you have a message like this: |
39 | 3 | Luke Murphey | |
40 | 3 | Luke Murphey | <pre> |
41 | 3 | Luke Murphey | Unable to initialize modular input "web_ping" defined inside the app "website_monitoring": Introspecting scheme=web_input: script running failed (exited with code 1). |
42 | 3 | Luke Murphey | </pre> |
43 | 3 | Luke Murphey | |
44 | 3 | Luke Murphey | You can search the internal log for related errors that indicate the modular input could not start: |
45 | 3 | Luke Murphey | |
46 | 3 | Luke Murphey | <pre> |
47 | 3 | Luke Murphey | index=_internal "ModularInputs - Unable to initialize modular input" |
48 | 3 | Luke Murphey | </pre> |
49 | 3 | Luke Murphey | |
50 | 3 | Luke Murphey | The best way to identify the source of the error if the modular input cannot be loaded by Splunk is to try to load it via Splunk's python interpreter directly. |
51 | 3 | Luke Murphey | |
52 | 3 | Luke Murphey | Here is an example. Consider the modular input is in the following path: /opt/splunk/etc/apps/website_monitoring/bin/web_ping.py |
53 | 3 | Luke Murphey | |
54 | 3 | Luke Murphey | To test this, go to the path /opt/splunk/etc/apps/website_monitoring/bin/ using the CLI: |
55 | 3 | Luke Murphey | |
56 | 3 | Luke Murphey | <pre> |
57 | 3 | Luke Murphey | cd /opt/splunk/etc/apps/website_monitoring/bin/ |
58 | 3 | Luke Murphey | </pre> |
59 | 3 | Luke Murphey | |
60 | 3 | Luke Murphey | Next, run Splunk's python against the file: |
61 | 3 | Luke Murphey | |
62 | 3 | Luke Murphey | <pre> |
63 | 3 | Luke Murphey | /opt/splunk/bin/splunk cmd python web_ping.py |
64 | 3 | Luke Murphey | </pre> |
65 | 3 | Luke Murphey | |
66 | 3 | Luke Murphey | This should show an error, such as: |
67 | 3 | Luke Murphey | |
68 | 3 | Luke Murphey | <pre> |
69 | 3 | Luke Murphey | Traceback (most recent call last): |
70 | 3 | Luke Murphey | File "web_input.py", line 1, in <module> |
71 | 3 | Luke Murphey | somethingnotdefined |
72 | 3 | Luke Murphey | NameError: name 'somethingnotdefined' is not defined |
73 | 3 | Luke Murphey | </pre> |
74 | 5 | Luke Murphey | |
75 | 5 | Luke Murphey | h2. Determining the cause of errors relating to "splunkd connection error" |
76 | 5 | Luke Murphey | |
77 | 5 | Luke Murphey | This happens when the input is running but Splunkd fails to respond when the input attempts to determine if a proxy server was configured. This appears to happen to Splunk installs that are running fine. In most cases that I can verify, it happens when Splunk has just been started. These can likely be safely ignored unless other issues have been observed that are related or if you see a large number of them. |
78 | 6 | Luke Murphey | |
79 | 6 | Luke Murphey | h2. Authentication is failing against sites that use Windows Authentication (a.k.a. NTLM authentication) |
80 | 6 | Luke Murphey | |
81 | 6 | Luke Murphey | Search the Website Monitoring logs for information that indicates which authentication type is being used. The app will attempt to discover the authentication type automatically. If this fails, then authentication will not succeed. |
82 | 6 | Luke Murphey | |
83 | 6 | Luke Murphey | <pre> |
84 | 6 | Luke Murphey | | search index=_internal sourcetype="web_availability_modular_input" auth_type |
85 | 6 | Luke Murphey | </pre> |
86 | 7 | Luke Murphey | |
87 | 7 | Luke Murphey | h2. Why are newly added URLs triggering continuously irrespective of Interval time? Or, how do I fix errors "Failed to save checkpoint directory"? |
88 | 7 | Luke Murphey | |
89 | 7 | Luke Murphey | This can happen when the app doesn't have access to the modular input checkpoint directory. Check the permissions of the modular input checkpoint directory (e.g. "C:\Program Files\Splunk\var\lib\splunk\modinputs\web_ping") to see if the account that Splunk runs under has read/write access to it. |
90 | 7 | Luke Murphey | |
91 | 7 | Luke Murphey | If it doesn't, edit the permissions. The input should begin functioning correctly as soon as the permissions are set correctly. |
92 | 8 | Luke Murphey | |
93 | 8 | Luke Murphey | h2. Some of my inputs that use authentication are not working and I am seeing the error "Unable to determine authentication type"? |
94 | 8 | Luke Murphey | |
95 | 8 | Luke Murphey | Websites can support several types of authentication. The input tries to determine what type of authentication the website supports. The error "Unable to determine authentication type" occurs when the input couldn't determine the type of authentication that the websites supports. When this happens, it will default to HTTP Basic authentication. |
96 | 10 | Luke Murphey | |
97 | 10 | Luke Murphey | h2. Some of my inputs are returning a 403 error even though I haven't told them to use authentication. |
98 | 10 | Luke Murphey | |
99 | 10 | Luke Murphey | This is likely because you have a proxy server configured but you input is unable to authenticate with the proxy. Oftentimes, this happens because there is a global HTTP proxy defined using using environment variables (often in $SPLUNK_HOME/etc/splunk-launch.conf). |
100 | 10 | Luke Murphey | |
101 | 10 | Luke Murphey | To get around this, define a value for the proxy_ignore setting in website_monitoring.conf. You can also set this using the setup page for the Website Monitoring app. Set the value of the "Server Ignore List" to "*", like this: |
102 | 10 | Luke Murphey | |
103 | 10 | Luke Murphey | !setup.png! |