On Nov 14, 2020, at 11:17, Nils Breunese wrote: > Ryan Schmidt wrote: > >> Sounds like maybe another consequence of the massive number of processes >> that MacPorts starts when trying to determine which of the hundreds of perl >> master_sites is closest to you. See: >> >> https://trac.macports.org/ticket/60509 > > I indeed saw about 123 ping processes running while trying to fetch > p5.28-locale-gettext. > >> We added some some code to mitigate that in MacPorts 2.6.3, reducing the >> number of processes by 2/3: >> >> https://github.com/macports/macports-base/commit/cee74bb45141be217449c04ee0acd4cb22042dbe >> >> We added further code to limit the number of processes to 50, but that has >> only been committed to master, so it is not currently scheduled to appear >> before MacPorts 2.7.0: >> >> https://github.com/macports/macports-base/commit/0a000c436ce0f8bc7da7bb34efad896dca6a1e59 >> >> Your log shows you're using MacPorts "2.6.4", which hasn't been released >> yet, so you must be using the release-2.6 branch. Try either adding the >> above commit to it or building master instead and see if that helps. > > I indeed installed MacPorts on Big Sur 11.0.1 via the latest state of the > release-2.6 branch, because the 2.6.3 release failed to build. > > I just installed MacPorts from master and indeed saw the number of ping > processes was limited to 50 now. But my machine still crashed after hanging > for 3 minutes on '---> gettext-1.07.tar.gz does not exist in > /opt/local/var/macports/distfiles/perl5’.
You could edit the source code to change 50 to a smaller number and try again. To ensure that your test is not influenced by previously cached entries, delete /opt/local/var/macports/pingtimes before trying again. If you didn't see this problem with other ports that are not perl modules, then presumably there is some smaller number of simultaneous pings that will work. But it also sounds like it may be a bug in the Carbon Black software you mentioned using.
