Interesting stuff, Matt. Note, though, that if the problem is in OpenSSH, then it won't apply to Net::SSH (or, by extension, Capistrano), since Net::SSH is pure Ruby and doesn't use any of the OpenSSH code.
I'm still suspecting Net::SSH, myself, and it's wild and unnecessary use of threads. The good news, though, is that Net::SSH v2 is coming along nicely, and uses IO.select instead of threads, which it should have done all along. - Jamis On 7/26/07, matt <[EMAIL PROTECTED]> wrote: > > I wonder if this SMP/OpenSSH bug could be the cause of the corrupted > mac issue. I get it a lot, as well as "Connection reset by peer" > errors when doing a deploy via checkout instead of via copy, so I'm > screwed no matter which way I turn :( It should be easy enough (?) to > test this out by running all remote commands through taskset, but I > don't know the cap internals well enough to try this myself. > > http://blog.mecworks.com/articles/2007/02/13/cpu-affinity-how-to-work-around-broken-ssh-on-smp-machines-using-taskset/ > > Matt > > > > > --~--~---------~--~----~------------~-------~--~----~ To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/capistrano -~----------~----~----~----~------~----~------~--~---
