I was testing errors in the cloud using juju and created a crash report for d-feet. The first time I did this the a .crash file was created and then a corresponding .upload and .uploaded file were created. The second time I did this a new .crash file was created but a new .upload file was not and subsequently the 2nd occurrence of this crash was not sent to the error tracker.
My /var/crash directory:
-rw-rw-r-- 1 bdmurray whoopsie 0 Nov 28 10:25 _usr_bin_d-feet.1000.upload
-rw------- 1 whoopsie whoopsie 0 Nov 28 13:14 _usr_bin_d-feet.1000.uploaded
-rw-r----- 1 bdmurray whoopsie 3401262 Nov 28 13:43 _usr_bin_d-feet.1000.crash
In apport/fileutils.py we can see that there are no checks done to see if there is already a .upload or .uploaded file:
I was testing errors in the cloud using juju and created a crash report for d-feet. The first time I did this the a .crash file was created and then a corresponding .upload and .uploaded file were created. The second time I did this a new .crash file was created but a new .upload file was not and subsequently the 2nd occurrence of this crash was not sent to the error tracker.
My /var/crash directory:
-rw-rw-r-- 1 bdmurray whoopsie 0 Nov 28 10:25 _usr_bin_ d-feet. 1000.upload d-feet. 1000.uploaded d-feet. 1000.crash
-rw------- 1 whoopsie whoopsie 0 Nov 28 13:14 _usr_bin_
-rw-r----- 1 bdmurray whoopsie 3401262 Nov 28 13:43 _usr_bin_
In apport/fileutils.py we can see that there are no checks done to see if there is already a .upload or .uploaded file:
def mark_report_ upload( report) :
report = '%s.upload' % report.rsplit('.', 1)[0]
with open(report, 'a'):
pass
So if a .upload for the crash already exists a new one is not created nor is the existing one modified.