Any reason to not let travis run #3960?
[qa] travis: cover more tests #7019
pull MarcoFalke wants to merge 1 commits into bitcoin:master from MarcoFalke:MarcoFalke-2015-receivedby changing 1 files +1 −1-
MarcoFalke commented at 9:43 PM on November 14, 2015: member
-
[qa] travis: cover *receivedby* rpcs 33b7f83c59
-
sipa commented at 9:54 PM on November 14, 2015: member
What is the reason for not running all of them in Travis? Are some too slow? Perhaps just one of the test combinations could run them all?
- MarcoFalke force-pushed on Nov 14, 2015
- MarcoFalke force-pushed on Nov 14, 2015
- MarcoFalke force-pushed on Nov 15, 2015
-
MarcoFalke commented at 12:14 AM on November 15, 2015: member
Are some too slow?
One uses 4 GB disk space and at least half an hour time: https://github.com/bitcoin/bitcoin/blob/master/qa/rpc-tests/pruning.py#L9-l11
In theory travis could handle those within the 120 minute time limit but I am not sure about other limits.
-
MarcoFalke commented at 12:33 AM on November 15, 2015: member
Travis appears to fail running the extended tests with
Warning! This test requires 4GB of disk space and takes over 30 mins (up to 2 hours) Mining a big blockchain of 995 blocks Unexpected exception caught during testing: [Errno 111] Connection refusedWhich is not due to a full disk, I guess. This would throw
Error: Error: Disk space is low! - MarcoFalke renamed this:
[trivial] travis: cover *receivedby* rpcs
[qa] travis: cover more tests
on Nov 15, 2015 -
jonasschnelli commented at 9:49 AM on November 15, 2015: contributor
Running the
-extendedtests during a CI build/check seems to be to expansive. Especiallypruning.pyuses a lot of resources (and time).The idea is to enables as much tests as possible over the non-extended tests list.
Enabling -extended in Travis seems against the concept. Adding more tests to the non-extended list should be the way to go.
The idea of the extended list is to have an option to run expansive tests locally (and it also serves as a overview of all available tests).
- jonasschnelli added the label Tests on Nov 15, 2015
- MarcoFalke force-pushed on Nov 15, 2015
-
MarcoFalke commented at 10:12 AM on November 15, 2015: member
@jonasschnelli : Reverted to inital commit.
-
jonasschnelli commented at 12:17 PM on November 15, 2015: contributor
ACK.
-
fanquake commented at 12:48 PM on November 15, 2015: member
utACK
On Sunday, November 15, 2015, Jonas Schnelli notifications@github.com wrote:
ACK.
— Reply to this email directly or view it on GitHub #7019 (comment).
-
jamesob commented at 5:30 AM on November 16, 2015: member
utACK
-
jamesob commented at 5:32 AM on November 16, 2015: member
I like the approach here -- was just about to dive into writing some additional RPC tests when I found that we have coverage in the extended tests for a few that're reported as uncovered. We should probably do an audit to figure out which of the -extended tests can reasonably be run on travis, then add some smoke tests for those that remain uncovered.
-
MarcoFalke commented at 12:13 PM on November 16, 2015: member
@jamesob I am working on this right now. (Needs some refactor, though)
-
laanwj commented at 7:50 AM on November 20, 2015: member
What is the reason for not running all of them in Travis? Are some too slow? Perhaps just one of the test combinations could run them all?
The problem is that a pull request check only finishes when all combinations are finished. CI is not very useful if you have to wait hours for a travis result to come in every time, so it relies on fast tests.
The idea of the extended list is to have an option to run expansive tests locally (and it also serves as a overview of all available tests).
I think it would be useful to run the extended tests, say, daily on master. No idea if Travis can be configured like that.
-
jonasschnelli commented at 7:55 AM on November 20, 2015: contributor
I think it would be useful to run the extended tests, say, daily on master. No idea if Travis can be configured like that.
I think there is a tools that can trigger a nightly build: https://nightli.es , although not sure if its easy to provide different arguments so that the build will execute the -extended tests.
There is also a discussion here: https://github.com/travis-ci/travis-ci/issues/582
-
laanwj commented at 8:00 AM on November 20, 2015: member
That looks useful!
By the way: ACK,
receivedby.pyseems to finish in 11 seconds when the cache is already there, a few seconds longer if not. So it's not close to problematic. - laanwj merged this on Nov 20, 2015
- laanwj closed this on Nov 20, 2015
- laanwj referenced this in commit 07b770caf3 on Nov 20, 2015
- MarcoFalke deleted the branch on Nov 30, 2015
- luke-jr referenced this in commit b529156021 on Jan 10, 2016
- luke-jr referenced this in commit 1407c705ab on Jan 10, 2016
- DrahtBot locked this on Sep 8, 2021