Cleanup nodes and jobs #15
Conversation
| @@ -68,7 +70,7 @@ def runParallel(args) { | |||
| /* We want to timeout a node if it doesn't respond | |||
| * The timeout should only start once it is acquired | |||
| */ | |||
| timeout(time: 60, unit: 'MINUTES') { | |||
| timeout(time: 3, unit: 'MINUTES') { | |||
There was a problem hiding this comment.
Why did this change drastically?
There was a problem hiding this comment.
Because we expect this to timeout if a reboot happens (the node disconnects and for some reason Jenkins doesn't notice it back online).
Since we are only running a quick on target test it should take no longer than 2 mins to flash, test, fail, basic powercycle, reflash, test, issue reboot command. 3 mins was chosen as to be more safe (as I only survey a few boards).
| @@ -95,12 +97,28 @@ def stepRunNodeTests() | |||
|
|
|||
| def stepUnstashBinaries(test) { | |||
| unstash name: "${env.BOARD}_${test.replace("/", "_")}" | |||
| sh script: "rsync -a tests/shell/ /opt/RIOT/tests/shell/" | |||
| sh script: "rsync -a ${TEST}/ /opt/RIOT/${TEST}/" | |||
There was a problem hiding this comment.
Just one example of unquoted variables which make me cringe. Can't this be 'rsync -a "${TEST}/" "/opt/RIOT/${TEST}/"'?
There was a problem hiding this comment.
Will cleanup... https://www.reddit.com/r/programminghorror/
There was a problem hiding this comment.
Ahhh I actually want the TEST variable defined in the groovy script to be resolved.
This should resolve into
> rsync -a tests/shell/ /opt/RIOT/tests/sched_testingdue to this
If I single quote it, it will actually run resolve TEST on the node and not in the groovy script. Since it doesn't exist in the node env it will just be empty.
Can we chalk this up to groovy interpreting unpleasantness?
There was a problem hiding this comment.
Fine, if no variables are passed to the shell it's fine
|
Ideally I figure out a way to have common functions factored out. This can be done later though as we see a huge improvement in reliability, agility and usability. |
MrKevinWeiss
force-pushed
the
pr/rack4cleanup
branch
from
d288a94
to
45d9674
Compare
|
So many more issues have been uncovered. It think first we need to get the fixes into RobotFW-tests then we can add these things. |
|
https://hil.riot-os.org/results/staging_rf_tests/1/overview show the most up to date result. On my cpu it seems like the periph_i2c is really big... Maybe it is unbound as all the other columns stay the same size but this one moves. ping @ozfox |
This does happen only on chromium based browsers because of their table rendering. The table is rather complicated because of the sticky header + sticky first column. It should be a grid anyway but is a table for historical reasons as I did not refactor it but only changed styles and did some bug fixing. I created a Issue for this but won't fix it right now. RIOT-OS/RobotFW-frontend#8 |
Currently on prod and things are running well.