Implementation of the token-safe retry logic for gfal #12191
Conversation
|
@khurtado FYI, we could start testing this in a testbed node instrumented with token authentication, while we extend the implementation to the |
| self.setups = "env -i X509_USER_PROXY=$X509_USER_PROXY JOBSTARTDIR=$JOBSTARTDIR bash -c '{}'" | ||
| elif auth_method == "TOKEN": | ||
| self.setups = "env -i BEARER_TOKEN=$(cat $BEARER_TOKEN_FILE) JOBSTARTDIR=$JOBSTARTDIR bash -c '{}'" | ||
| else: |
There was a problem hiding this comment.
@amaltaro do we want to allow gfal-cp to run even when no authentication method is specified?
| except StageOutError as ex: | ||
| msg = "Attempt {} to stage out failed.\n".format(retryCount) | ||
| msg = "Attempt {} to stage out failed with default setup.\n".format(retryCount) |
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
test this |
|
test this please |
|
Jenkins results:
|
|
@amaltaro I am not sure that the new failing unit test is due to some changes of mine |
|
Jenkins results:
|
|
These are the newly failing unit tests: which I am not able to find under the @d-ylee @khurtado am I doing something wrong? How do I get to the details of the failing unit tests? Do you understand why the 2 reported error/failures don't show up in the list of 51 failing unit tests? |
|
@amaltaro This is interesting. The error shows up in 117, but not in 118. https://cmssdt.cern.ch/dmwm-jenkins/job/WMCore-PR-Report/117/#showFailuresLink Looking at the GitHub comment history, it looks like both you and @anpicci asked Jenkins to do the test and also made a new commit at around the same time, so I am assuming 117 and 118 are from both of your comments. |
|
test this please |
|
Jenkins results:
|
|
retest this please |
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
|
Jenkins results:
|
Fixes #12144
Status
not-tested
Description
This PR introduces the retry logic proposed by @stlammel for handling the possible failures with token authentication when used with
gfal-cp. To be extended to thexrootdprotocol.More details in the issue description
Is it backward compatible (if not, which system it affects?)
YES
Related PRs
None
External dependencies / deployment changes
None