Sorry for the rather vague subject-line, but I can''t think of any pithy way to describe what I''m seeing. I have Puppet 0.24.4 running on a whole bunch of Solaris machines. Some 10, some 8, some SPARC, some x86. On two of them (8/SPARC, but not the only 8/SPARC machines we have!) this: file {"/etc/resolv.conf": source => [ "puppet:///dist/resolv.conf.$hostname", "puppet:///dist/resolv.conf.$location", "puppet:///dist/resolv.conf" ] } Results in "/etc/resolv.conf" becoming a symlink to "/etc/resolv.conf.mel" rather than simply containing the contents of puppet:///dist/resolv.conf.mel. ($location is set by a custom fact.) There are other seemingly-identical Solaris 8/SPARC systems running the same Puppet client and Ruby build (it''s all rsynced from one common copy) in the same machine room which work properly, and it worked properly on these hosts with our previous version of Puppet (0.24.1). Anyone else seen anything like this? It''s a bit disconcerting that a "file" stanza which specifies a source and does not have an "ensure" clause which would create a link winds up creating a link. Matt --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Puppet Users" group. To post to this group, send email to puppet-users@googlegroups.com To unsubscribe from this group, send email to puppet-users-unsubscribe@googlegroups.com For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en -~----------~----~----~----~------~----~------~--~---
On May 1, 2008, at 12:18 AM, Matt McLeod wrote:> There are other seemingly-identical Solaris 8/SPARC systems running > the same > Puppet client and Ruby build (it''s all rsynced from one common copy) > in the > same machine room which work properly, and it worked properly on > these hosts > with our previous version of Puppet (0.24.1). > > Anyone else seen anything like this? It''s a bit disconcerting that > a "file" > stanza which specifies a source and does not have an "ensure" clause > which > would create a link winds up creating a link.The only time I''ve seen anything like this is when the http keepalive was causing file corruption. You''re absolutely sure those broken clients are running 0.24.4? I just can''t imagine this discrepancy in behaviour without some distinct configuration difference. -- Don''t throw away the old bucket until you know whether the new one holds water. -- Swedish Proverb --------------------------------------------------------------------- Luke Kanies | http://reductivelabs.com | http://madstop.com --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Puppet Users" group. To post to this group, send email to puppet-users@googlegroups.com To unsubscribe from this group, send email to puppet-users-unsubscribe@googlegroups.com For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en -~----------~----~----~----~------~----~------~--~---
Luke wrote:> The only time I''ve seen anything like this is when the http keepalive > was causing file corruption. > > You''re absolutely sure those broken clients are running 0.24.4?Yup. I have two builds of a tree containing the Ruby stable-snapshot from a few days ago, Puppet 0.24.4, and whatever the latest Facter is (both installed as gems). One is built on Solaris 8/SPARC (and running on both 8 and 10 on SPARC), the other on Solaris 10/x86. The hosts that are misbehaving are even configured identically in Puppet to others that aren''t -- they''re all nodes inheriting the same "meta-node". Apologies for the crappy quoting and probably screwed-up wordwrap/etc, I''m stuck using Outlook/Entourage.> I just can''t imagine this discrepancy in behaviour without some > distinct configuration difference.Looking more closely at the OS patch revisions on the two sets of hosts, it seems that the misbehaving ones have slightly older (relatively speaking, this is Solaris 8 after all!) versions of a few patches that may be relevant: the kernel, libc/libc++, and linker patches. Not sure I can actually patch those as these are largely "hands off" machines, but I''ll see and report back. Matt --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Puppet Users" group. To post to this group, send email to puppet-users@googlegroups.com To unsubscribe from this group, send email to puppet-users-unsubscribe@googlegroups.com For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en -~----------~----~----~----~------~----~------~--~---
On May 2, 2008, at 1:37 AM, McLeod, Matt wrote:> > Luke wrote: >> The only time I''ve seen anything like this is when the http keepalive >> was causing file corruption. >> >> You''re absolutely sure those broken clients are running 0.24.4? > > Yup. I have two builds of a tree containing the Ruby stable-snapshot > from a few days ago, Puppet 0.24.4, and whatever the latest Facter is > (both installed as gems). One is built on Solaris 8/SPARC (and > running > on both 8 and 10 on SPARC), the other on Solaris 10/x86. > > The hosts that are misbehaving are even configured identically in > Puppet > to others that aren''t -- they''re all nodes inheriting the same > "meta-node". > > Apologies for the crappy quoting and probably screwed-up wordwrap/etc, > I''m stuck using Outlook/Entourage. > >> I just can''t imagine this discrepancy in behaviour without some >> distinct configuration difference. > > Looking more closely at the OS patch revisions on the two sets of > hosts, > it seems that the misbehaving ones have slightly older (relatively > speaking, this is Solaris 8 after all!) versions of a few patches that > may be relevant: the kernel, libc/libc++, and linker patches. Not > sure > I can actually patch those as these are largely "hands off" machines, > but I''ll see and report back.I think this is beyond my abilities; you''ve got near-identical hosts that behave differently. While it''s clearly possible that this is a Puppet bug, and given the behaviour it''s maybe even likely, but... how could I hope to track it down? -- Sometimes I think we''re alone. Sometimes I think we''re not. In either case, the thought is staggering. --R. Buckminster Fuller --------------------------------------------------------------------- Luke Kanies | http://reductivelabs.com | http://madstop.com --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Puppet Users" group. To post to this group, send email to puppet-users@googlegroups.com To unsubscribe from this group, send email to puppet-users-unsubscribe@googlegroups.com For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en -~----------~----~----~----~------~----~------~--~---