sequential 4 seems to delay log output
Closed this issue · 1 comments
the promote action seems to output it's logs delayed when using :sequential: 4
in cvmanager.yaml
steps to reproduce:
-
use a cvmanager (git pull-ed 2017-09-12)
-
set
:sequential: 4
incvmanager.yaml
-
have some (about 18 in this case) CCV to promote defined in
cvmanager.yaml
-
have
time ./cvmanager --wait promote --to-lifecycle-environment 30
in a wrapper script (cvauto.sh
also callstime ./cvmanager --wait publish
andtime ./cvmanager --wait update
plus some other non-cvmanager commands) -
/root/katello-cvmanager/cvauto.sh | tee /root/katello-cvmanager/cvauto_$(date +"%Y-%m-%d_%H-%M-%S").log
-
try to follow a Satellite task in the webUI by gettng it's ID from the cvmanager output
actual result:
Tasks get scheduled and executed in batches of 4 as intended.
But I can follow Synchronize repository …
Tasks in the Satellite webUI way before they show up in the shell. Some may finish before I see them in the cvmanager output
e.g. 2df778d3-e553-4c36-8bc1-befd0a36bac8
ran as follows:
Started at: 2017-09-12 18:56:40 UTC
Ended at: 2017-09-12 19:11:04 UTC
but in the tee
I saw it only some time after 19:16 UTC
What I get is long period of silence and when there is output it's about many tasks followed by another long (double digit minutes) period of silence.
expected result:
seeing output like the following around the time the tasks are queued, not many minutes later.
Inspecting ccv-demo-rhel6-qpid
Promoting latest version to lifecycle-environment 30
waiting 10 for pending tasks: ["b552cc77-173b-4ea9-b4d2-27d882dd42bf"]
waiting 20 for pending tasks: ["b552cc77-173b-4ea9-b4d2-27d882dd42bf"]
addition info:
- Thanks for adding the sequential option, I find it extremely useful!
- Other output from the script show up in the tee fluid as expected.
- To me this delayed output does not matter much, normally the wrapper script runs at night from cron as
/root/katello-cvmanager/cvauto.sh > /root/katello-cvmanager/cvauto_$(date +"%Y-%m-%d_%H-%M-%S").log 2>&1
and the logs are looked at in the morning, but I figured you'd want to know. - I see the same delayed log chunks on the clean action. Did not notice on other actions (e.g. update) yet, but then again I did not attentively follow the other actions called by the wrapper script.
- the clean action triggered, amongst others, "5c70e2f1-8443-4a30-8191-9b72443952f4" which ran from 19:23:32 UTC to 19:23:36 UTC yet shell output about the task was only some time after 19:27, with many other clean tasks in the same log burst (that had mostly also already finished).
- the terminal I ran this in remains perfectly usable, so it's not my connection. ;-) Maybe I'm just using tee wrong.