Some packages get "killed" on the Linux builders during a daily run

[hpages]

Some packages get "killed" on the Linux builders during the daily builds. See https://stat.ethz.ch/pipermail/bioc-devel/2024-February/020167.html for the details.

Here is the list of all software packages that got "killed" during the last run (Feb 7, 2024)...

.. of the BioC 3.19 builds:

• https://bioconductor.org/checkResults/3.19/bioc-LATEST/multiGSEA/nebbiolo1-checksrc.html
• https://bioconductor.org/checkResults/3.19/bioc-LATEST/MungeSumstats/nebbiolo1-checksrc.html
• https://bioconductor.org/checkResults/3.19/bioc-LATEST/netDx/nebbiolo1-checksrc.html
• https://bioconductor.org/checkResults/3.19/bioc-LATEST/omicRexposome/nebbiolo1-checksrc.html
• https://bioconductor.org/checkResults/3.19/bioc-LATEST/OMICsPCA/nebbiolo1-checksrc.html
• https://bioconductor.org/checkResults/3.19/bioc-LATEST/ontoProc/nebbiolo1-checksrc.html
• https://bioconductor.org/checkResults/3.19/bioc-LATEST/orthos/nebbiolo1-checksrc.html

.. of the BioC 3.18 builds:

• https://bioconductor.org/checkResults/3.18/bioc-LATEST/MungeSumstats/nebbiolo2-checksrc.html
• https://bioconductor.org/checkResults/3.18/bioc-LATEST/octad/nebbiolo2-checksrc.html
• https://bioconductor.org/checkResults/3.18/bioc-LATEST/omicRexposome/nebbiolo2-checksrc.html

Here are the commands that were used to obtain the above lists (showing the commands for 3.19):

wget https://bioconductor.org/checkResults/3.19/bioc-LATEST/report.tgz
tar zxf report.tgz
cd report
grep Killed */nebbiolo1-*.html
# multiGSEA/nebbiolo1-checksrc.html:Killed
# MungeSumstats/nebbiolo1-checksrc.html:  Killed
# MungeSumstats/nebbiolo1-checksrc.html:Killed
# netDx/nebbiolo1-checksrc.html:Killed
# omicRexposome/nebbiolo1-checksrc.html:Killed
# OMICsPCA/nebbiolo1-checksrc.html:Killed
# ontoProc/nebbiolo1-checksrc.html:Killed
# orthos/nebbiolo1-checksrc.html:Killed

To get the list of packages:

grep Killed */nebbiolo1-*.html | cut -d '/' -f 1 | sort | uniq
# multiGSEA
# MungeSumstats
# netDx
# omicRexposome
# OMICsPCA
# ontoProc
# orthos

These kills only seem to happen during the CHECK stage (R CMD check).

One thing we could try is reduce the number of workers used during that stage in order to reduce the load on the machine. The nebbiolos are very fast builders and they currently finish a daily run hours before the other machines. Slightly reducing the number of workers (e.g. from 28 to 24) will slightly increase the time it takes for the machines to complete a run but that's something that we can afford. We just need to make sure that they don't start to overlap with other builds that start early in the morning e.g. it would be a problem if the machines now finish around 3 am instead of 1 am and some other builds start around 3 am.

H.

P.S.: It seems that the magick package is needed to build omicRexposome's vignette so omicRexposome should put magick in its Suggests field. @lshep Maybe something for your team of helpers to report to the omicRexposome people?

[hpages]

I made that change (see d3df597). Looks like the nebbiolos ususally finish their run between 2:30 am and 3:30 am. The earlier thing that runs in the morning are the data-annotation builds which start at 5 am (on Wednesday only). I changed them to start at 6 am.