Better end of console log indexing detection.
* elastic_recheck/query_builder.py: Due to the way multiline events work in logstash, the "Finished: FAILURE" line is not pushed into elasticsearch until the next console.html file is processed or some timeout is reached (I haven't actually confirmed that the timeout works though). This happens because this is the last line in the file that we do not drop. To better detect the end of indexing look for the "[SCP] Copying console log" message which should also happen at the end of each console log file. A better long term solution to this problem may be to have the logstash gearman workers insert and EOF event that can ensure each log file is completely flushed without waiting for more data. Change-Id: I7c51e019c774f9ba15e731aca01aaffc27de1e60
This commit is contained in:
parent
f05111a794
commit
604b6468ee
@ -66,7 +66,9 @@ def result_ready(review=None, patch=None):
|
||||
lets us know that we've got results waiting that we need to process.
|
||||
"""
|
||||
return generic('filename:"console.html" AND '
|
||||
'(@message:"Finished: FAILURE" OR message:"Finished: FAILURE") '
|
||||
'(@message:"[SCP] Copying console log" '
|
||||
'OR message:"[SCP] Copying console log") '
|
||||
'AND build_status:"FAILURE" '
|
||||
'AND build_change:"%s" '
|
||||
'AND build_patchset:"%s"' %
|
||||
(review, patch))
|
||||
|
Loading…
x
Reference in New Issue
Block a user