On Mon, 03 Sep 2018 22:23:18 +0100, Mick wrote: > On Monday, 3 September 2018 19:10:52 BST Holger Hoffstätte wrote: >> On Mon, 03 Sep 2018 17:48:53 +0100, Mick wrote: >> > I just noticed today chronyd hangs during boot for a minute and a half. >> > The logs do not reveal anything amiss. I suspect it waits for a network >> > connection, which is not yet up when chronyd launches. >> > >> > Sep 3 17:38:26 asus syslog-ng[1858]: syslog-ng starting up; >> > version='3.13.2' Sep 3 17:38:26 asus chronyd[1930]: chronyd version 3.3 >> > starting (+CMDMON +NTP +REFCLOCK +RTC -PRIVDROP +SCFILTER -SIGND >> > +ASYNCDNS -SECHASH +IPV6 -DEBUG) Sep 3 17:38:26 asus chronyd[1930]: >> > Frequency -13.776 +/- 1.023 ppm read from /var/lib/chrony/drift >> > Sep 3 17:38:26 asus acpid[1894]: starting up with netlink and the input >> > layer Sep 3 17:38:26 asus acpid[1894]: 1 rule loaded >> > Sep 3 17:38:26 asus acpid[1894]: waiting for events: event logging is off >> > Sep 3 17:40:57 asus kernel: random: crng init done >> > Sep 3 17:40:57 asus kernel: random: 3 urandom warning(s) missed due to >> > ratelimiting >> > Sep 3 17:40:57 asus chronyd[1930]: Loaded dump file for 10.10.10.1 >> > >> > >> > Have you noticed the same? >> > >> > Is there a fix/workaround for this? >> >> Yes and yes; countless people have been hit by this, I was probably the >> first to notice/debug it. It happens due to changes in Linux' random number >> generator, which is now slowly making it into older kernels as well. >> >> If you're interested in the gory details you can read about them >> in the thread "Chrony vs. Linux RNG" at [1] or just update to chrony-3.4rc1 >> which contains the "fix". Despite being an rc1 it works fine as usual, >> so short of applying the patch to your local build this is the easiest >> option. >> >> cheers, >> Holger >> >> [1] >> https://listengine.tuxfamily.org/chrony.tuxfamily.org/chrony-users/2018/04/ >> threads.html#00036 > > Thanks for this Holger, I had noticed in the logs a message just before > chronyd started the kernel reporting "3 urandom warning(s) missed due to > ratelimiting" but as this has been happening since the upgrade to 4.14.65- > gentoo I didn't think it was kernel related. > > Anyway, I can confirm 3.4_pre1 available in the tree starts up fine with no > delays. > > However, while shutting down I noticed: > > * Stopping chronyd ... > * Error: chronyd failed to stop > > This is repeatable if I try to stop it manually. Version 3.3 didn't have > this > problem.
Ah yes, that's from upgrading. Apologies for the misleading advice, I fixed that locally and forgot. :} 1) Edit /etc/init.d/chronyd and change occurrences of /run/chronyd.pid to /run/chrony/chronyd.pid (i.e. add the subdirectory). 2) file a bug so that the init file gets fixed in portage as well. :) cheers Holger
