Stream: committers
Topic: CI-Build Failure
Grahame Grieve (Jun 19 2019 at 19:15):
@Josh Mandel the build seems to have failed completely right now, and it's a big problem. There are two problems:
- US-Core-R4 has been rebuilt plenty recently, but the summary page shows it's last rebuild as more than a month ago
- the build started failing completely overnight. It doesn't look to me like it's my part
Grahame Grieve (Jun 19 2019 at 19:15):
here's a log:
Grahame Grieve (Jun 19 2019 at 19:16):
INFO:root:about to clone!
DEBUG:root:running: git clone --recursive https://github.com/esource-consortium/fhir-clinical-research --branch master repo
DEBUG:root:Time to deadline 1679.996293 seconds
Cloning into 'repo'...
DEBUG:root:running: wget -q https://github.com/FHIR/latest-ig-publisher/raw/master/org.hl7.fhir.publisher.jar -O publisher.jar
DEBUG:root:Time to deadline 1678.774739 seconds
DEBUG:root:running: java -Xms4550m -Xmx4550m -jar ../publisher.jar -ig ig.json -auto-ig-build -target https://build.fhir.org/ig/esource-consortium/fhir-clinical-research/ -out /scratch/ig-build-temp-NS3CM8/repo
DEBUG:root:Time to deadline 1676.84361 seconds
FHIR Implementation Guide Publisher Version 0.9.37-SNAPSHOT - Built 2019-06-19T22:13:33.882+10:00 - Git c8dbdc48a05f
Detected Java version: 12.0.1 from /usr/java/openjdk-12 on amd64 (64bit). 4398MB available
Run time = Wednesday, June 19, 2019 at 4:59:40 PM Coordinated Universal Time
[/scratch/ig-build-temp-NS3CM8/repo] -ig ig.json -auto-ig-build -target https://build.fhir.org/ig/esource-consortium/fhir-clinical-research/ -out /scratch/ig-build-temp-NS3CM8/repo
Package Cache: var/lib/.fhir/packages
Load Configuration from /scratch/ig-build-temp-NS3CM8/repo/ig.json (00.0038sec)
Redirecting to Configuration from /scratch/ig-build-temp-NS3CM8/repo/src-generated/cr-core-ig.json (00.0057sec)
Publishing Content Failed: /scratch/ig-build-temp-NS3CM8/repo/src-generated/cr-core-ig.json (No such file or directory) (00.0057sec)
(00.0057sec)
Use -? to get command line help (00.0057sec)
(00.0057sec)
Stack Dump (for debugging): (00.0058sec)
java.io.FileNotFoundException: /scratch/ig-build-temp-NS3CM8/repo/src-generated/cr-core-ig.json (No such file or directory)
at java.base/java.io.FileInputStream.open0(Native Method)
at java.base/java.io.FileInputStream.open(FileInputStream.java:213)
at java.base/java.io.FileInputStream.<init>(FileInputStream.java:155)
at org.hl7.fhir.utilities.TextFile.fileToString(TextFile.java:153)
at org.hl7.fhir.utilities.json.JsonTrackingParser.parseJsonFile(JsonTrackingParser.java:362)
at org.hl7.fhir.igtools.publisher.Publisher.initializeFromJson(Publisher.java:1464)
at org.hl7.fhir.igtools.publisher.Publisher.initialize(Publisher.java:1119)
at org.hl7.fhir.igtools.publisher.Publisher.execute(Publisher.java:603)
at org.hl7.fhir.igtools.publisher.Publisher.main(Publisher.java:6115)
Exception in thread "main" java.lang.NullPointerException
at org.hl7.fhir.igtools.publisher.Publisher.main(Publisher.java:6125)
Josh Mandel (Jun 19 2019 at 19:17):
I'm at the airport but have some bandwidth to debug. Let's leave aside the IG Build summary page for now -- doesn't seem urgent unless you think it'd somehow related
Grahame Grieve (Jun 19 2019 at 19:17):
no the second is more urgent.
Josh Mandel (Jun 19 2019 at 19:18):
Okay. My first debugging step will be to see if I get the same issue when I run the IG build docker container locally. Does that sound right?
Grahame Grieve (Jun 19 2019 at 19:18):
no that might actually be genuine error.
Grahame Grieve (Jun 19 2019 at 19:18):
I get that one locally
Grahame Grieve (Jun 19 2019 at 19:20):
ok sorry. The first 3 build failures I picked were all the one guide and it's that guide that's failing
Grahame Grieve (Jun 19 2019 at 19:20):
the other guides are failing because of me. I am investigating...
Grahame Grieve (Jun 19 2019 at 19:20):
so back to the first issue - the status file being out of date - thanks
Josh Mandel (Jun 19 2019 at 19:21):
OK -- status file out-of-date isn't shocking given the changes we made for supporting multiple branches. Is it just happening for US-Core-R4 ?
Grahame Grieve (Jun 19 2019 at 19:22):
no looks selective, if you cross-compare the notifications channel with the status file
Josh Mandel (Jun 19 2019 at 19:29):
Just put in a fix -- it's rendering a timestamp from an older branch instead of the master branch.
Josh Mandel (Jun 19 2019 at 19:30):
(Still fixing, actually ;-))
Josh Mandel (Jun 19 2019 at 19:33):
Anyway, no urgent issues in the build itself, right @Grahame Grieve ?
Josh Mandel (Jun 19 2019 at 19:33):
(Date issue in https://fhir.github.io/auto-ig-builder/builds.html is fixed now:
https://github.com/FHIR/auto-ig-builder/commit/fb835a17ea6892962bd798a86b6d99de3ea5e673 for details)
Grahame Grieve (Jun 19 2019 at 19:34):
it's still broken for me in Chrome :-(
Grahame Grieve (Jun 19 2019 at 19:35):
do you see failing guides listed? there's a bunch of davinci guides I'm not seeing listed at all because their last build failed
Josh Mandel (Jun 19 2019 at 19:35):
I'm also using Chrome; can you try ctrl-shift-reload? If that fails, try again in a bit -- GH Pages uses a CDN that can take time to propagate changes.
Grahame Grieve (Jun 19 2019 at 19:35):
working now, but not seeing the failing builds
Josh Mandel (Jun 19 2019 at 19:36):
Ooh, interesting -- no, I'm not seeing failed DaVinci guides. that's a separate (not a new as of 2 minutes ago) issue, right?
Grahame Grieve (Jun 19 2019 at 19:37):
I hadn't seen it before but don't know when it became a problem
Josh Mandel (Jun 19 2019 at 19:38):
OK -- will look at that next.
Josh Mandel (Jun 19 2019 at 20:08):
When builds fail, we have to go parsing through the build.log to figure out things like when, and which version. It looks like the format of files like https://build.fhir.org/ig/HL7/davinci-pdex-formulary/branches/master/build.log has changed at som point, so dates are written like
Run time = Wednesday, June 19, 2019 at 6:16:19 PM Coordinated Universal Time
which doesn't match the parsing logic (" @ {{ISO-8601 date}}}")
Josh Mandel (Jun 19 2019 at 20:09):
That's part of the issue at least.
Josh Mandel (Jun 19 2019 at 20:19):
I've added a bit of defense against this, so the failed guides now show up (again) -- but they won't have a proper date attached to them.
Grahame Grieve (Jun 19 2019 at 20:29):
@Josh Mandel so this is ok:
Run time = Thursday, 20 June 2019 6:28:39 AM AEST (@2019-06-20T06:28:39+10:00)
Grahame Grieve (Jun 19 2019 at 20:29):
?
Josh Mandel (Jun 19 2019 at 21:06):
Technically you want @ (that is, spaces around the at symbol), per my fragile parsing code ;-)
Josh Mandel (Jun 19 2019 at 21:07):
Obviously I should fix that
Grahame Grieve (Jun 19 2019 at 21:07):
ok
Run time = Thursday, 20 June 2019 7:07:25 AM AEST ( @ 2019-06-20T07:07:25+10:00)
Grahame Grieve (Jun 19 2019 at 21:07):
I'll release that now
Josh Mandel (Jun 19 2019 at 21:09):
Then we can try rebuilding the davinci guides.
Josh Mandel (Jun 19 2019 at 21:10):
And they should get dates (or more likely, I can debug why they don't)
Grahame Grieve (Jun 19 2019 at 21:19):
ok released
Last updated: Apr 12 2022 at 19:14 UTC
