Custom Fingerprinting Rules
If our default grouping algorithm is separating occurrences that you would rather have combined or vice versa, then you can set up Custom Fingerprinting Rules.
To set your custom grouping rules, go to Project Settings -> [Your project] -> Custom Fingerprinting for the project you want to configure.
Here's an example configuration:
[
{"condition": {"path": "body.trace.exception.class", "eq": "TimeoutError"},
"fingerprint": "timeout-error",
"title": "Timeout Error"
},
{"condition": {"path": "body.trace.exception.class", "eq": "IndexError"},
"fingerprint": "index-error"
}
]
The above configuration is a list of rules. Each rule is a JSON object that consists of a condition, a fingerprint, and an optional title.
Rules are applied in order, testing the condition against the incoming occurrence. If a match is found, the rule's fingerprint and title will be used instead of our default algorithm.
This only applies when your SDK isn't also sending an explicit fingerprint in the payload. If it is, the explicit fingerprint takes precedence over your rule by default for the vast majority of projects — see Precedence below.
Occurrences with the same fingerprint are combined into an Item.
The title of the first occurrence of an Item is used as the title of the Item, and is only changed if the item is resolved and then later reactivated. Upon reactivation, the title of the reactivating occurrence is used as the new Item title.
Condition
The condition is defined as a single JSON object. It can either be:
- a leaf condition, having a
pathkey and an operator key (see list below). - a branch condition, having a single key named
any,all, ornone, whose value is a list of other conditions.
The path key describes a path from the root of the occurrence data to the data that the condition tests. Use dots (.) to select object properties or array elements. You can also use the * character in place of an array index or property name to select all matching elements; in this case, the condition will pass if the operator evaluates to true for any of the matching elements.
Path
Here are a few commonly used paths:
| Path | Description |
|---|---|
body.trace.exception.class | Exception class name |
body.trace.exception.message | Exception message |
body.trace.frames.0.filename | Filename of the first stack frame |
body.trace.frames.-1.filename | Filename of the last stack frame |
body.trace.frames.*.filename | Filename of any stack frame |
body.trace.frames.*.method | Method/function name of any stack frame |
language | The language name |
You can use any value in your JSON payload as a path, including custom data. To view your JSON payloads, go into an occurrence and click on View JSON at the bottom of the screen.
Here's a simple example of some error JSON:
{
"body": {
"trace": {
"frames": [
{
"method": "HTMLDocument.e._wrapped",
"lineno": 32,
"colno": 100,
"filename": "https://cdn.rollbar.com/rollbarjs/refs/tags/v2.15.0/rollbar.min.js"
}
],
"exception": {
"message": "Boomerang is not defined",
"class": "ReferenceError",
"description": "Error while initializing Heroku header"
}
}
},
"server": {
"host": "web01"
}
}
In this example, the value of body.trace.frames.0.lineno is 32, and the value of server.host is web01.
Operators
The following operators are available:
| Operator | Description |
|---|---|
eq | Equals |
neq | Not Equals |
in | Contained in the string or list |
nin | Not contained in the string or list |
contains | Contains the string, element, or key (case sensitive) |
icontains | Contains the string, element, or key (case insensitive) |
ncontains | Does not contain the string, element, or key (or is not a string, array, or object) |
startswith | Starts with the string |
endswith | Ends with the string |
gt | Greater than |
gte | Greater than or equal to |
lt | Less than |
lte | Less than or equal to |
regex_match | Regular expression matcher |
For the numeric operators (gt, gte, lt, lte): the right side must be a number, and the left side (data from the occurrence) will be coerced to a number, defaulting to 0.
Example: Match items whose exception class is NameError
{
"path": "body.trace.exception.class",
"eq": "NameError"
}
Example: Match items whose exception class is TypeError or ValueError, and exception message contains the string "database"
{
"all": [
{ "any": [
{ "path": "body.trace.exception.class", "eq": "TypeError" },
{ "path": "body.trace.exception.class", "eq": "ValueError" }
]},
{ "path": "body.trace.exception.message", "contains": "database" }
]
}
Note: If your exception has nested stack traces, rather than using body.trace.exception.message, you'll need to use body.trace_chain.0.exception.message, and so on for any paths that begin with body.trace.
Example: Match items whose message body is 7 to 10 digits and does not begin with 4321
{
"all": [
{"path": "body.message.body", "regex_match": "^(?!4321).*"},
{"path": "body.message.body", "regex_match": "^\\d{7,10}$"}
]
}
Fingerprint
Occurrences with the same fingerprint are combined into an Item. To add occurences to an existing group item created by manual merging, use the fingerprint group-item-###, e.g. group-item-123. See the merging guide for an example.
Title
The title is an optional text description that is displayed when viewing an item. It must be a string, of length 1-255 characters. You can change the title in this configuration without impacting the fingerprint. The new title will take effect if the item is reactivated after being resolved.
Templates
Both fingerprint and title can contain template markers (wrapped in {{double braces}}) as well as hardcoded strings.
There are two special markers:
{{default_fingerprint}}: will be replaced with the fingerprint hash calculated by our default algorithm.{{default_title}}: will be replaced with the title calculated by our default algorithm.
These can be used to tune the fingerprinting algorithm without entirely replacing it.
Additionally, any part of the occurrence JSON body may be referenced by path. For example
{{ body.trace.exception.class }}will be replaced with the exception class.{{ body.trace.exception.message }}will be replaced with the exception message.
You can combine this with static text or the special markers achieve many kinds of fingerprinting.
Complete Example
Here's an example complete configuration with three rules.
[
{
"condition": {
"path": "body.trace.exception.class",
"in": ["EOFError", "Errno::ECONNREFUSED", "Errno::ETIMEDOUT"]
},
"fingerprint": "connection-error",
"title": "Connection error"
},
{
"condition": {"all": [
{"path": "body.trace.exception.class", "eq": "TimeoutError"},
{"path": "body.trace.frames.*.filename", "contains": "payments"}
]},
"fingerprint": "payments-timeout-error",
"title": "Timeout error in payments code"
},
{
"condition": {
"path": "body.trace.exception.class",
"eq": "ActionController::RoutingError"
},
"fingerprint": "{{ default_fingerprint }}-{{ context }}",
"title": "{{default_title }} in {{ context }}"
}
]
Precedence
If your SDK sends an explicit fingerprint in an occurrence's payload, that fingerprint takes precedence over your custom grouping rules by default: the matching rule is skipped, and the item is grouped on the explicit fingerprint instead. This is the default behavior for every project created after June 26, 2017, which is the vast majority of Rollbar projects. Only projects created on or before that date default the other way, with custom rules taking precedence over an explicit fingerprint.
If you've written a rule that never seems to match, and you're also sending an explicit fingerprint from your SDK, this default is almost always why.
To make your custom grouping rules take precedence over an explicit SDK fingerprint instead, go to Project Settings -> [Your project] -> Custom Fingerprinting and check the 'Ignore Client-side fingerprinting' box. With that box checked, Rollbar evaluates your rules first and only falls back to the SDK's fingerprint if no rule matches.
Updated 17 days ago