Thanks for clearing that up for me Walter. There have been so many people with the same error I mixed your name up with someone else. As you may know, taking down a nameserver to implement these changes is not a trivial matter for larger enterprises. I am very disappointed that the response to your problem was that "the product was not fully matured yet and if you need stability go with 9.4.2-P1". I still have yet to meet the person who "doesn't" need stability in the DNS.
I have seen so many threads of problems with so many of the post-cache-bug versions of BIND that I doubtful that any of them are safe to use. I know that the ISC people had to rush these products out because of the cache problem, so I don't fault them, but I have to wonder if the cure is worse than the disease? I don't mean to bitch about things, but I just don't have time to chase down possible remedies that are unproven. I hope that a concrete cause can be found, so that we can make concrete plans for addressing the issue. If it's an actual sizing limitation - tell us. if it's a thread problem - tell us. Just do the due diligence and get it right so we can nail the issue. Emery, On Fri, Jul 25, 2008 at 11:01 AM, Walter Gould <[EMAIL PROTECTED]> wrote: > Emery Rudolph wrote: > >> Milan, >> After getting more information I see that the 65535 limit I was seeing for >> the FD_SETSIZE limit is strictly for 64-bit applications. I will try the >> recommended path of setting the 32-bit value to 4096 and recompiling BIND. >> >> The one thing that bothers me is that Walter Gould (an earlier contributor >> to this thread) is also running Solaris 9, has the exact same error and >> tried the recommended modification and recompile only to wind up with >> continued crashes. I will not make too strong a correlation, because there >> are other variables such as system hardware that could be the cause. >> >> >> > Emery, > Actually, I am running an older version of Red Hat Enterprise - not Solaris > 9. However, I have had problems with receiving the "Too man open files" > error. When this happens, the named daemon continues to run - but does not > resolve names. Also, I have recompiled named increasing the FD_SETSIZE to > 4096. Weird thing is after doing that, named ran for about a day before I > received the "Too man open files" error and named stopped resolving. > Walter > >
