Jenkins Pipeline - Reading previous stage log - jenkins

Question

Consider a Jenkins Pipeline with two stages, Stage A then Stage B.
In Stage B, is it possible to parse the logs of Stage A for some particular text?

Answer 1

Use tee to split the output to both stdout and file. Next parse the file for your text.
STAGE_A_LOG_FILE = 'stage_a.log'
pipeline {
agent any
stages {
stage('Stage A') {
steps {
script {
// tee log into file
tee(STAGE_A_LOG_FILE) {
echo 'print some Stage_A log content ...'
}
}
}
}
stage('Stage B') {
steps {
script {
// search log file for 'Stage_A'
regex = java.util.regex.Pattern.compile('some (Stage_A) log')
matcher = regex.matcher(readFile(STAGE_A_LOG_FILE))
if (matcher.find()) {
echo "found: ${matcher.group(1)}"
}
}
}
}
}
}
Pipeline output:
print some Stage_A log content ...
found: Stage_A
Finished: SUCCESS

Answer 2

There's been an update since July 28th !
As mentionned in this answer, as of version 2.4 of Pipeline: Nodes and Processes you can use:
def out = sh script: 'command', returnStdout: true
At least it's much more simple and clean than outputting to a file and reading the file afterwards.

Answer 3

What I finally did was, as suggested, to write to a file (and stdout) using tee.
sh "command | tee <filename>"
Then parse the file as needed, using readFile <filename> to read the file from the workspace.

Answer 4

If you want to search for the first occurrance of a pattern, you can also use manager.logContains(regexp) or manager.getLogMatcher(regexp). See my other answer for more details: https://stackoverflow.com/a/39873765/4527766